View Full Version : RV560/570 Gemini roadmap
Some benchmark numbers....
1950XTX512 - 11000 3DM05
1950PRO256 - 9558 3DM05 / 5320 3DM06
1650XT256 - 6690 3DM05 / 3724 3DM06
1650PRO512 - 5269 3DM05 / 2943 3DM06
1650PRO256M - 5269 3DM05 / 2943 3DM06
Here are the kickers...
X1950XTX wil be cheaper than a GF7900GTX
X1950Pro will be cheaper than a GF7600GT OC
X1650XT will be cheaper than a GF7600GT std
X1650Pro512 will cost around the same as a GF7600GS512 OC
X1650Pro256 will cost a bit more than a GF7600GS std
It looks like A(Ti)MD is going for a pricewar again.
Chalnoth
02-Aug-2006, 02:21
It looks like A(Ti)MD is going for a pricewar again.
That's something that nVidia can handle better than ATI with their current architecture.
Was that ATI or AMD/ATI you meant with that, Chal? :smile:
IgnorancePersonified
02-Aug-2006, 02:33
Well they are already in a cpu price war why not take one up the challenge on the gpu front ?
X1950Pro (RV570) looks very tempting. :smile:
Anon Lamer
02-Aug-2006, 05:32
Will there be 512 meg skus for the 1950pro/1650XT? In f.ex BF2 that would make a difference.
asicnewbie
02-Aug-2006, 05:49
while thinking about it I've found a solution for the ugly DVI problem. you know that DB9, DB25, VGA, DVI connectors have those screws so they don't move. So, in the future when I have a DVI-I only card, instead of screwing the DVI adapter into the card, I'll screw the VGA cord into the DVI adapter : it will thus be part of the VGA cord, both psychologically and pratically :grin:
(yes, still deeper PC but I won't have to think about the DVI dongle when moving it..)
If the separate 'DVI-VGA dongle' bothers you that much, you could always buy a DVI-A to DB-15 cable -- no separate dongle-needed!
Thanks, CJ. And to your source(s). :)
1950PRO256 - 9558 3DM05 / 5320 3DM06
1650XT256 - 6690 3DM05 / 3724 3DM06
[...]
X1950Pro will be cheaper than a GF7600GT OC
X1650XT will be cheaper than a GF7600GT std
You lost me there. Did you mean 7900GT OC vs. X1950P?
To put things in perspective, according to TR's 3DM06 #s in their 7950GX2 review (w/ an X2 4800+):
7675 7950GX2
5950 7900GTX
5865 X1900XT
4588 X1800XT
4514 7900GT
And 3DM05 / 06 #s courtesy of the three latest XbitLabs reviews (FX-60; it's possible the X1900XT is a repeated typo and actually 5902, given TR's #):
10877 / 6040 7900GTX
10709 / 5092 X1900XT
9509 / 4560 X1900GT
9373 / 4625 X1800XT
8013 / 4538 7900GT
7572 / 3762 X1800XL
6906 / 3275 X1800GTO
6060 / 3319 7600GT
Chalnoth
02-Aug-2006, 09:15
Was that ATI or AMD/ATI you meant with that, Chal? :smile:
Well, it's still ATI for the time being.
That's something that nVidia can handle better than ATI with their current architecture.
But AIB partners could evaluate that ATi's solutions need cheaper memory modules to stay competitive (e.g. X1900XTX 1550MHz vs. 7900GTX 1700MHz).
Thanks, CJ. And to your source(s). :)
You lost me there. Did you mean 7900GT OC vs. X1950P?
No. I meant what I wrote... GF7600GT OC. :) X1900XT256 is going up against the 7900GT.
To put things in perspective, according to TR's 3DM06 #s in their 7950GX2 review (w/ an X2 4800+):
7675 7950GX2
5950 7900GTX
5865 X1900XT
4588 X1800XT
4514 7900GT
And 3DM05 / 06 #s courtesy of the three latest XbitLabs reviews (FX-60; it's possible the X1900XT is a repeated typo and actually 5902, given TR's #):
10877 / 6040 7900GTX
10709 / 5092 X1900XT
9509 / 4560 X1900GT
9373 / 4625 X1800XT
8013 / 4538 7900GT
7572 / 3762 X1800XL
6906 / 3275 X1800GTO
6060 / 3319 7600GT
What I didn't mention was that the slide also contained comparison numbers for all the nVidia cards that the ATi cards are competing against. And it's definitely X1950Pro vs GF7600GT OC, because the nVidia card is listed as scoring 6042 in 3DM05 and 3745 in 3DM06. So it's definitely X1950Pro vs GF7600GTOC.
anyone kind enough to post translation? ie what number (x1950xtxsuper), wwhat chip is (r582+?) and what was its previous name (like x1650ProLELE was x1600XTX)
10x :twisted:
X1950 XTX 512MB - $449 (R580+)
X1900 XTX 512MB - $349 (R580)
X1900 XT 512MB - $299 (R580)
X1900 XT 256MB - $249 (R580)
X1950 Pro 256MB - $199 (RV570)
X1650 XT 256MB - $149 (RV560)
X1650 Pro 512MB - $129 (RV530)
X1650 Pro 256MB - $99 (RV530)
My guesses based on what CJ said. ATI could also discontinue the 512MB sku of the X1900 XT as it looks pointless in the stack.
CJ, did they cancel the XL version of the RV570? Or is it coming later (Dec?): http://i6.tinypic.com/15316vd.jpg
+edited to include gpu codenames.
chavvdarrr
02-Aug-2006, 17:45
anyone kind enough to post translation? ie what number (x1950xtxsuper), wwhat chip is (r582+?) and what was its previous name (like x1650ProLELE was x1600XTX)
10x :twisted:
X1950 XTX 512MB - $449 (R580+)
X1900 XTX 512MB - $349 (R580)
X1900 XT 512MB - $299 (R580)
X1900 XT 256MB - $249 (R580)
X1950 Pro 256MB - $199 (RV570)
X1650 XT 256MB - $149 (RV560)
X1650 Pro 512MB - $129 (RV530)
X1650 Pro 256MB - $99 (RV530)
My guesses based on what CJ said. ATI could also discontinue the 512MB sku of the X1900 XT as it looks pointless in the stack.
CJ, did they cancel the XL version of the RV570? Or is it coming later (Dec?): http://i6.tinypic.com/15316vd.jpg
+edited to include gpu codenames.
ATi's roadmaps change a lot. What would you expect from a RV570XL? Would it be a RV570XT just clocked slower or do you think they've disabled a quad (turning it into a 256-bit version of the RV560)? And most of all... In what price range would you put it if you see the pricelist that you've just posted (which btw is close to the truth)?
BTW, X1650Pro will make use of RV535XT when they've ran out of RV530XT cores. And all RV530Pro and RV535Pro cores will be used in X1300XT ($79-$99).
CJ, that's a hyoooge peformance jump at MSRP$200. Very nice. Also, I forgot the Pro was RV570, I thought it was the same as the GT: a cut-down R580. Whoops. :)
serenity, seems like an equally huge jump going from a $200 RV570 to a $250 R580. I'd be surprised if performance tracked so well with price (33% > performance for just 25% > price?!) I'd expect the current XTX to go the way of the dodo, leaving a $279-299 X1900XT 256 and maybe a $349 X1900XT 512. The X1950XT should stay at $450+.
The jump from RV530 to RV560 also seems implausibly large for the price. Practically twice the card for just 50% more?
But I guess the market will be different when RV570 & 560 are released in October. so that tight and seemingly low pricing may well be accurate. I won't stand in the way of progress!
ATi's roadmaps change a lot. What would you expect from a RV570XL? Would it be a RV570XT just clocked slower or do you think they've disabled a quad (turning it into a 256-bit version of the RV560)? And most of all... In what price range would you put it if you see the pricelist that you've just posted (which btw is close to the truth)?
Ah, makes sense. It'll be interesting to see how they will make use of the borked RV570s (one quad less). I would have spread some green to you but stupid forum rules dont allow me to. :-x
serenity, seems like an equally huge jump going from a $200 RV570 to a $250 R580. I'd be surprised if performance tracked so well with price (33% > performance for just 25% > price?!) I'd expect the current XTX to go the way of the dodo, leaving a $279-299 X1900XT 256 and maybe a $349 X1900XT 512. The X1950XT should stay at $450+.
The jump from RV530 to RV560 also seems implausibly large for the price. Practically twice the card for just 50% more?
But I guess the market will be different when RV570 & 560 are released in October. so that tight and seemingly low pricing may well be accurate. I won't stand in the way of progress!
Honestly after a long time it looks like ATI is bringing goodness to the sub $300 and $200 price-points. I wonder how Nvidia will counter these new parts, their G8x range wont be there so they'll have to make use of current ones.
Chalnoth
02-Aug-2006, 22:00
Honestly after a long time it looks like ATI is bringing goodness to the sub $300 and $200 price-points. I wonder how Nvidia will counter these new parts, their G8x range wont be there so they'll have to make use of current ones.
The GeForce 7900GT is currently as low as about $250, so they don't have far to go in price to compete well with the Radeon X1950 Pro.
The GeForce 7900GT is currently as low as about $250, so they don't have far to go in price to compete well with the Radeon X1950 Pro.
That is a MSRP v SRP comparison. Also ATI is revamping sub $200 categories as well.
Chalnoth
02-Aug-2006, 22:39
That is a MSRP v SRP comparison. Also ATI is revamping sub $200 categories as well.
Right. I'm just saying that it wouldn't be all that much of a price drop to remain competitive.
And we all know that street prices can end up above MSRP's just as well as they can end up below them.
trumphsiao
03-Aug-2006, 09:24
anyone kind enough to post translation? ie what number (x1950xtxsuper), wwhat chip is (r582+?) and what was its previous name (like x1650ProLELE was x1600XTX)
10x :twisted:
from what I perceived Die Size of Rv570 is 2xxmm^2(less than 65% of R580):razz:
also RV5XX series have some problem need to be unhinged
1.Less TMU than G7X series from top to bottom
2. PCB Cost
Pretty good guess (http://www.newegg.com/Product/Product.asp?Item=N82E16814142068) about RV570, serenity.
trumphsiao, was it that cheap since its release, or is that a recent price drop? How expensive is, say, a 7900GT or an X1900XT, for comparison?
trumphsiao
03-Aug-2006, 09:33
Pretty good guess (http://www.newegg.com/Product/Product.asp?Item=N82E16814142068) about RV570, serenity.
down here in Taiwan you can come with 200USD to purchase a X1900GT:wink:
Blazkowicz
03-Aug-2006, 22:41
The jump from RV530 to RV560 also seems implausibly large for the price. Practically twice the card for just 50% more?
it's a bit like what was geforce 6600 vs 6600GT (where GPU was much faster clocked, sure not twice, but the bandwith was almost double)
the more low-end you get the more cards stink compared to their price anyway, a 6200 is much worse than a 7300GS, which is much worse than 7300GT or 7600GS.
trumphsiao
04-Aug-2006, 10:13
it's a bit like what was geforce 6600 vs 6600GT (where GPU was much faster clocked, sure not twice, but the bandwith was almost double)
the more low-end you get the more cards stink compared to their price anyway, a 6200 is much worse than a 7300GS, which is much worse than 7300GT or 7600GS.
RV5XX series line-up is ATI the most advancable but less margin compared to GF7 series. 7900GT consignment price is less than 250~230USD .when ATI put RV560/570 on the shelves of global market as a whole. the less period of time ATI put stockpile of chips onto the AIB channel.the less margin of loss is remained to be seen.
we all concur that G80/R600 is "water-cool" motif design design form.but as I perceived from the preliminary G80 sample. I dont feel any thing engrossing .
Chalnoth
04-Aug-2006, 12:15
Water cooling is still going to be a very niche market with the next generation of GPU's.
Some news about RV550:
Friends from graphic market said ATi plans to merge Xilleon graphic processor into Radeon GPU, targeting on HTPC and Digital Home market. The merge enables ATi’s value and mainstream products to be capable for 1080p AVIVO HD Video Decoding, which is essential to the upcoming Blu-Ray and HD-DVD.
Xilleon is designed for digital TVBox as a Universal Video Decoder inside digital TV. It’s a 300MHz 32Bit MIPS processor. ATi plans to merge it into its GPU to enable 40Mbps + CABAC Blu-Ray decoding without occupying GPU’s resources. Further support in Dual-Stream decoding, 1080p HD Video decoding is no longer a problem for mainstream platforms...
http://www.hkepc.com/bbs/news.php?tid=641493
http://img208.imageshack.us/img208/7663/uvd1yhmg2k2ho5kdjn5.th.jpg (http://img208.imageshack.us/my.php?image=uvd1yhmg2k2ho5kdjn5.jpg) http://img208.imageshack.us/img208/3884/uvd2kmqaiaxf7o0vaq1.th.jpg (http://img208.imageshack.us/my.php?image=uvd2kmqaiaxf7o0vaq1.jpg)
Sunrise
04-Aug-2006, 15:42
Some news about RV550
http://www.beyond3d.com/forum/showthread.php?t=32377 ;)
http://www.beyond3d.com/forum/showthread.php?t=32377 ;)
I was 5 days earlier than that. ;)
http://www.beyond3d.com/forum/showpost.php?p=798528&postcount=234
And thats why I posted these slides here :smile:
Sunrise
04-Aug-2006, 16:52
And thats why I posted these slides here :smile:
So it´s all geo´s fault, right ? ;)
Your post doesn´t have anything new, the date just changed to 8/4, that´s why i posted the link. CJ´s round-up, if a little earlier than that, was incomplete and he was also refering to an older news (http://www.hkepc.com/bbs/viewthread.php?tid=626532&extra=page%3D2) that HKEPC already posted about a month ago.
So it´s all geo´s fault, right ? ;)
No, of course. It's my fault. I should read whole B3Df before posting anything :grin:
Looks like the X1800 GTO² is coming. :smile:
http://i5.tinypic.com/23mqyis.jpg
An X1800XL with an XT cooler, any guesses on pricing?
Anon Lamer
05-Aug-2006, 07:26
Looks like the X1800 GTO² is coming. :smile:
An X1800XL with an XT cooler, any guesses on pricing?
Seek an older thread in the "gfx boards and drivers" forum.
Seek an older thread in the "gfx boards and drivers" forum.
This (http://www.beyond3d.com/forum/showthread.php?t=29040) one?
Looks like the X1800 GTO² is coming. :smile:
An X1800XL with an XT cooler, any guesses on pricing?
GTO2 is X1800XT-512 under-clocked to 500/500 (XL). It's already sold out according to our suppliers.
GTO2 is X1800XT-512 under-clocked to 500/500 (XL). It's already sold out according to our suppliers.
I wasnt aware of that. :oops: At what price?
The price was about 20-30% higher than basic X1800GTO. It was available only for 2 weeks (Czech Republic).
Anon Lamer
05-Aug-2006, 10:29
This (http://www.beyond3d.com/forum/showthread.php?t=29040) one?
Sigh now you made me go looking for it. Its this one: http://www.beyond3d.com/forum/showthread.php?t=31671
The GeForce 7900GT is currently as low as about $250, so they don't have far to go in price to compete well with the Radeon X1950 Pro.
The 7900GT isn't going to drop in price... Welcome GeForce 7950GT 512. Same price as the GeForce 7900GT 256.
But ATi will have a little counter against the GF7950GT 512 shortly after the introduction of the X1950XTX.
trumphsiao
18-Aug-2006, 17:28
http://www.ocer.net/article/a_upload/200608/20060817-1650.gif
woo, 12X3 ALUs :grin: , RV570= 256bit RV560 version:shock:
trumphsiao
18-Aug-2006, 17:55
RV570 12X3 ALUs 12 TMUs 256bit
RV560 12X3 ALUs 12TMUs 128bit
:twisted: but BOM cost is too high, sigh...............
RV570= 256bit RV560 version:shock:
Are you reading off the Nvidia slides or is there another source? :lol:
$199 for RV570 is sweet. It should easily fight off 7900GS.
INKster
18-Aug-2006, 21:54
Are you reading off the Nvidia slides or is there another source? :lol:
$199 for RV570 is sweet. It should easily fight off 7900GS.
I'm not so sure...
12 vs 20 ?
INKster
18-Aug-2006, 22:54
According to PC INpact (http://www.pcinpact.com/actu/news/30819-Quelles-seront-les-performances-de-la-procha.htm), the 7950 GT (in this case, a "simulated one, using an overclocked 7900 GT) is up to 21% faster than a regular 7900 GT, and only 9% slower than a full blown X1900 XT.
Is RV570 faster than a X1900 XT, or at least faster than a 7950 GT ?
At $299 ?
I'll choose the best of the two.
Dave Glue
18-Aug-2006, 23:26
According to PC INpact (http://www.pcinpact.com/actu/news/30819-Quelles-seront-les-performances-de-la-procha.htm), the 7950 GT (in this case, a "simulated one, using an overclocked 7900 GT) is up to 21% faster than a regular 7900 GT, and only 9% slower than a full blown X1900 XT.
How about when running AA with HDR?
Oh...:)
I'm not so sure...
12 vs 20 ?
RV570 should be identical in performance to the X1900GT. In comparison 7900GS will be slower.
According to PC INpact (http://www.pcinpact.com/actu/news/30819-Quelles-seront-les-performances-de-la-procha.htm), the 7950 GT (in this case, a "simulated one, using an overclocked 7900 GT) is up to 21% faster than a regular 7900 GT, and only 9% slower than a full blown X1900 XT.
Is RV570 faster than a X1900 XT, or at least faster than a 7950 GT ?
At $299 ?
I'll choose the best of the two.
CJ hinted that the new X1900XT 256MB will be faster than a 7950GT 512MB.
X1900XT 512MB > X1900XT 256MB > 7950GT 512MB > X1950 Pro > 7900GS > RV560 > 7600GT*
*Guess
INKster
19-Aug-2006, 01:08
How about when running AA with HDR?
Oh...:)
I really don't care about that.
No card in this category has the speed for such quality settings (yet) anyway.
As for AA+HDR, i'll stick to "HL2: Ep1".
I prefer high resolution in games to having AA enabled on lower resolution settings, and that for a long time now.
But this is just too confusing, because i haven't seen a direct comparison between RV570 and the 7950 GT and i prefer to avoid the usual unsubstanciated hype.
Considering all the G80 details and iminent release, it might be wise for me to save some more money and buy one in a couple of months...:???:
But this is just too confusing, because i haven't seen a direct comparison between RV570 and the 7950 GT...
Why to compare these cards? X1950PRO(RV570) should be cheaper (http://www.beyond3d.com/forum/showpost.php?p=802881&postcount=251). Shouldn't it be X1950PRO x 7900GS and 7950GT x X1900XT-256?
Acert93
19-Aug-2006, 09:15
I really don't care about that.
No card in this category has the speed for such quality settings (yet) anyway.
As for AA+HDR, i'll stick to "HL2: Ep1".
I prefer high resolution in games to having AA enabled on lower resolution settings, and that for a long time now.
A number of recent benchmarks have shown AA + HDR on the X1800/X1900 series is not quite bad in regards to performance penalty.
But this is just too confusing, because i haven't seen a direct comparison between RV570 and the 7950 GT
X1950 Pro will be $200, 7950 GT will be $300. They are not competitors.
You can get the X1900 XT 512MB from online us retailers right now for around $300 so I don't personally see how the 7950 GT is any kind of deal.
INKster
19-Aug-2006, 19:57
X1950 Pro will be $200, 7950 GT will be $300. They are not competitors.
You can get the X1900 XT 512MB from online us retailers right now for around $300 so I don't personally see how the 7950 GT is any kind of deal.
X1950 Pro is the RV560, not the RV570, i think.
Anyway, i've decided to wait for DX10 cards.
Now i want leaks about R600 and G80 to decide this for good. :D
I think...
RV560 - X1650XT
RV570 - X1950PRO
I think...
RV560 - X1650XT
RV570 - X1950PRO
Yes (http://www.beyond3d.com/forum/showpost.php?p=797296&postcount=214). :smile:
INKster
19-Aug-2006, 21:59
I think...
RV560 - X1650XT
RV570 - X1950PRO
Yep, my mistake. :D
trumphsiao
20-Aug-2006, 13:34
Die size of RV560 is a lot bigger than that of 90nm G73.
Sunrise
20-Aug-2006, 13:48
Die size of RV560 is a lot bigger than that of 90nm G73.
Which is to be expected, since RV530 already is a fair bit larger than G73.
INKster
21-Aug-2006, 14:26
Which is to be expected, since RV530 already is a fair bit larger than G73.
And will be even larger, now that G73 was shrunk to 80nm:
http://www.hkepc.com/bbs/itnews.php?tid=655680
Sunrise
21-Aug-2006, 15:25
And will be even larger, now that G73 was shrunk to 80nm
Indeed. The margins on those will probably go through the roof (not if they weren´t already...). However, we have to wait and see how good the supply situation will be then.
RV570 12X3 ALUs 12 TMUs 256bit
RV560 12X3 ALUs 12TMUs 128bit
:twisted: but BOM cost is too high, sigh...............
RV560 is not 12x3. It's 8x3. It has 24 pixelshaderengines and works at 600Mhz core and 1400Mhz GDDR3 128-bit mem-interface.
DailyTech have some RV570 info and slides here, (http://www.dailytech.com/article.aspx?newsid=3973) together with the emerging product stack.
DailyTech have some RV570 info and slides here, (http://www.dailytech.com/article.aspx?newsid=3973) together with the emerging product stack.
Nice. My stack is realised. :cool:
On a side note, are these (new) slide designs a result of the aquisition?
Well, without specific ATI or AMD corporate branding present, it could be that the slides are either generic/temporary internal templates that DailyTech have simply watermarked while any branding issues in this regard are worked through in the merger, or they could perhaps be DailyTech created slides based on other ATI/AIB correspondence.
Something I can't understand is why ATI have so far neglected the X1700 nomenclature for RV560, instead relegating it to being marketed as X1650XT, which given the massive (speculated) performance difference between X1650 Pro (RV530/RV535), i.e. 8-1-3-1/2 in RV560 versus 4-1-3-2 in RV530/5, is going to confuse a lot of casual buyers and possibly even some less attentive enthusiasts.
Perhaps X1700 is being kept for a mobile RV560? Keeping the X1700 moniker for a spring refresh (e.g. RV560/RV570 at 65nm) wouldn't really jive, as the leading edge of the stack would have likely transitioned to the X2K range or similar by then and ATI tend to market products with the prevailing nomenclature, even if the featureset is a generation old, e.g. Radeon 9200, X600, X550, etc.
Anon Lamer
29-Aug-2006, 16:36
There should be a rv570 512 meg sku, there are always a 512 meg "pro" sku. Does the amount of memory that a card carry get decided at board maker level or does ATI decide that?
There should be a rv570 512 meg sku, there are always a 512 meg "pro" sku.
I believe I've seen a 512MB version of the RV570 being mentioned somewhere. It should cost about $239-249.
Anon Lamer
30-Aug-2006, 17:59
Nowadays staff at the stINQ doesnt even bother to search B3d before writning nonsense...
http://www.theinquirer.net/default.aspx?article=34025
Regarding the Dailytech slides, what is this "high speed anti-aliasing support"?
**Hopes for 4AA sample per clock rops**
Regarding the Dailytech slides, what is this "high speed anti-aliasing support"?
**Hopes for 4AA sample per clock rops**
Dailytech just made their own version of the official ATI slides. The original slides mention "Full Speed SuperAA" which DailyTech wrongly 'translates' to "high speed anti-aliasing support".
trumphsiao
04-Sep-2006, 12:41
http://www.ithlj.com/news/2006-9/25390_5.html
http://www.computerdiy.com.tw/modules/news/article.php?storyid=2360
INKster
04-Sep-2006, 13:00
That second link is a little weird.
So now Radeon X1650 becomes X1700, and Geforce 7650 becomes 7700 ?
Dave Baumann
04-Sep-2006, 13:12
Althought there's a reasonable correlation, branding for mobile doesn't necessarily follow the desktop naming schemes.
overclocked
04-Sep-2006, 14:12
Is both the 560 and 570 on 80nm (UMC/TSMC?) and do both share double Z-rate in the ROPs? Estimated die-sizes Dave?
...and do both share double Z-rate in the ROPs?
I think this could be an answer:
http://www.beyond3d.com/forum/showpost.php?p=819183&postcount=8
http://www.ithlj.com/news/2006-9/25390_5.html
http://www.computerdiy.com.tw/modules/news/article.php?storyid=2360
Some additional info about the MR X1700 can be found online. MR X1700 has the codename M66Pro and M66XT according to this site (http://pci-ids.ucw.cz/iii/?i=1002). According to the ATi site (http://www.ati.com/developer/vendorid.html) it's based on RV530. So it looks like it's a faster clocked MR X1600. Other MRs on that page are MR X1350 (M52 just like X1300) and MR X1450 (M54 just like X1400).
Dave Baumann
05-Sep-2006, 03:59
Is both the 560 and 570 on 80nm (UMC/TSMC?) and do both share double Z-rate in the ROPs? Estimated die-sizes Dave?
LOL! :lol:
X1950PRO pictured:
http://regmedia.co.uk/2006/09/05/ati_x1950_pro.jpg
http://www.reghardware.co.uk/2006/09/05/ati_x1950_pro_pics/
overclocked
05-Sep-2006, 20:01
LOL! :lol:
And while we are at it app release for R600(specs and all of curse), scheme for derivates and a big fat PM full of NDAs.. ;)
X1950PRO pictured:
http://regmedia.co.uk/2006/09/05/ati_x1950_pro.jpg
http://www.reghardware.co.uk/2006/09/05/ati_x1950_pro_pics/
Linky no worky, maybe you have them in your cache. :cool:
chavvdarrr
05-Sep-2006, 20:18
And while we are at it app release for R600(specs and all of curse), scheme for derivates and a big fat PM full of NDAs.. ;)
you forgot to demand hardware sample ;D
Linky no worky, maybe you have them in your cache. :cool:
The article was removed... But I saved it preventively :-)
http://img205.imageshack.us/img205/7426/atix1950proji4.th.jpg (http://img205.imageshack.us/my.php?image=atix1950proji4.jpg)
http://img205.imageshack.us/img205/7788/atix1950propwmrd2.th.jpg (http://img205.imageshack.us/my.php?image=atix1950propwmrd2.jpg)
http://img236.imageshack.us/img236/4871/atix1950procfcdk6.th.jpg (http://img236.imageshack.us/my.php?image=atix1950procfcdk6.jpg)
I'd post full verion, but I'm not sure if it wouln't infringe the rules of B3Df...
Man, that is one seriously clean VRM design.:shock:
http://img310.imageshack.us/img310/8846/rv570cfcm7.jpg
The sexy bridge. :smile:
I'm loving the new coolers. Thanks for the pics. :smile:
What is the other unused bridge for?
What is the other unused bridge for?
3rd, 4th, etc. cards?
Jawed
Good guess or shrewd tip off? :lol2:
:lol: Teehee - that mobo looks like it might have 3x PEGx16 slots!
Jawed
3rd, 4th, etc. cards?
Jawed
Doh, didnt notice it. :oops: Thats RD600 then.
Is it? My current i975X board has 3 PEG slots. I've just finished testing an i975X board with 4. Slot count doesn't indicate core logic ;)
Is it? My current i975X board has 3 PEG slots. I've just finished testing an i975X board with 4. Slot count doesn't indicate core logic ;)
Now you're just toying with me. :runaway:
Doh, didnt notice it. :oops: Thats RD600 then.
It's the DFI RD600 board.
trumphsiao
10-Sep-2006, 16:04
http://www.pcdvd.com.tw/showthread.php?t=651843
http://kelin.myweb.hinet.net/DSC00060.JPG
http://kelin.myweb.hinet.net/DSC00069.JPG
Skrying
10-Sep-2006, 16:11
I'm loving the new coolers. Thanks for the pics. :smile:
What is the other unused bridge for?
Second bridge connector possibly for it to interact with a third card for physics?
Nice, so this new X1900GT has almost the same PCB as X1950PRO - only w/o support for internal CF bridge.
Skrying: there is no CF connector on this card
I think probably a dedicated bridge is not needed as X1900GT are allowed to make CF via PCI-e 16 already (isn't it?). The second thing is that the chip might not have internal bridge hardware, thus cut-it-off would be a save bet for anyone confusion :wink:
Skrying
10-Sep-2006, 16:36
Nice, so this new X1900GT has almost the same PCB as X1950PRO - only w/o support for internal CF bridge.
Skrying: there is no CF connector on this card
Refering to the X1950 Pro, sorry didnt think of anyone to quote, so I see how that is confusing.
INKster
10-Sep-2006, 17:27
The core looks way too big to be a RV570, although i might be mislead by that huge chunk of thermal paste.
It IS R580 (A22 or A31) because the pictured card is X1900GT (the new 512/1320 HDCP revision) :-)
INKster
10-Sep-2006, 18:05
It IS R580 (A22 or A31) because the pictured card is X1900GT (the new 512/1320 HDCP revision) :-)
If that's the case, what is it doing in this thread ?
This was supposed to be the RV560/RV570 place...;)
If that's the case, what is it doing in this thread ?
Confusing? :-)
It seems like something is going to happen on the 17th of October.
vertex_shader
14-Sep-2006, 11:58
It seems like something is going to happen on the 17th of October.
Thats not good, rv560/570 already late :roll:
Daliytech has some news: http://www.dailytech.com/article.aspx?newsid=4166
Reference card running 580mhz core (rumors say 600mhz before) and 1400mhz memory clock, not so great yields? or ATi make some room for the AIB OC cards? (my dream is HIS iceq cooling x1950pro with 640/1550 clock speed, and for 199euro :wink:
HDCP optional, this is good, i not want pay someting what can't use for 2010-2012 :smile:
It seems Dailytech saw the same document that also mentions the 17th of October. ;) But somehow they forgot to mention the date and are vague about it... lol, I wonder why they just didn't just say it out loud. Oh well, the cat is out of the bag now. :P
Sunrise
14-Sep-2006, 16:38
It seems Dailytech saw the same document that also mentions...
Oh, did your Source tell you that? :lol::wink:
Fuad has heard about the 17th too....
http://uk.theinquirer.net/?article=34401
This time he corrects his mistake that he made in this article (http://uk.theinquirer.net/?article=34025) in which he wrote that RV560 was a RV530 on 80nm with 12 pixelshader engines. He corrected it to 24 pixelshader engines. :P Unfortunately he this time makes the mistake of calling RV570 'the new X1900GT' instead of X1950Pro.
chavvdarrr
15-Sep-2006, 11:02
Unfortunately he this time makes the mistake of calling RV570 'the new X1900GT' instead of X1950Pro.
eh, is there anyone in his mind, able to tell the difference between all kinds of x1300/1600/1650/1800/1900/1950's combined with se/pro/xt/gt/gto/OMFG or 6100/6150/6200/6600/6800/7300/7600/7800/7900 with tc/le/gs/gt/gtx/gx2/pleasehelpmeimgoinginsane ?!
trumphsiao
15-Sep-2006, 14:14
http://www.hkepc.com/bbs/news.php?tid=669300
R580 and RV570 are equally same die size.
vertex_shader
15-Sep-2006, 14:54
http://www.hkepc.com/bbs/news.php?tid=669300
R580 and RV570 are equally same die size.
http://www.hkepc.com/bbs/attachments_dir/ext_jpg/rv570_UGfOPeTW72yT.jpg
http://www.techpowerup.com/img/05-12-28/r580-1_LGn20UmqToIL.jpg
Rv570 looks much smaller than the r580.
trumphsiao
15-Sep-2006, 14:58
http://www.hkepc.com/bbs/attachments_dir/ext_jpg/rv570_UGfOPeTW72yT.jpg
http://www.techpowerup.com/img/05-12-28/r580-1_LGn20UmqToIL.jpg
Rv570 looks much smaller than the r580.
Chinese Article did state RV570 is almost as big as R580.
eh, is there anyone in his mind, able to tell the difference between all kinds of x1300/1600/1650/1800/1900/1950's combined with se/pro/xt/gt/gto/OMFG or 6100/6150/6200/6600/6800/7300/7600/7800/7900 with tc/le/gs/gt/gtx/gx2/pleasehelpmeimgoinginsane ?!
IIRC:
(PS/TMU/ROP)
X1300 4/4/4
X1600 12/4/4
X1650 12/4/4
These have regardless of SE/WTF/WhatEver the same amount of each unit)
X1800 16/16/16 (regardless of the SE/blahbla)
X1900GT 36/16/12 (in practice, 36/12/12)
X1950 Pro 36/12/12
X1900 XT 48/16/16
X1900 XTX 48/16/16
X1950 XTX 48/16/16
All the differences apart from the ones mentioned here, are in clocks and/or memory amounts.
Oh, and I think 1300-1600 have 128bit membus, while X1800 and up 256bit
vertex_shader
15-Sep-2006, 15:16
Chinese Article did state RV570 is almost as big as R580.
80nm value chip have the same size as a 90nm highend chip? sounds weird, maybe they need new glasses (or me? :wink:) , when i look this 2 picture, rv570 looks much smaller.
Normalising both pictures:
http://www.cupidity.f9.co.uk/b3d59.jpg
http://www.cupidity.f9.co.uk/b3d60.jpg
and measuring area, RV570 is 67% of the area of R580.
Naively, that would put RV570 at 260M transistors.
Jawed
Dave Baumann
15-Sep-2006, 15:26
IIRC:
(PS/TMU/ROP)
X1300 4/4/4
X1600 12/4/4
X1650 12/4/4...
Nearly. X1300 XT is 12/4/4! ;)
vertex_shader
15-Sep-2006, 15:39
"Diamond Multimedia (www.diamondmm.com), a leading manufacturer of PC graphics cards, sound cards and communications products, will begin shipping three new versions of its popular Viper video card line on September 14, including the X1950 CrossFire™ & X1950XTX in PCI-E models along with the X1950PRO available in an AGP version."
"Diamond Ships Six New Viper Cards
The X1950AGP model has a 256MB GDDR4 interface, core clock speed of 600MHz and a 1.4 GHz memory speed (Due Oct 2006). All these products support Dual DVI (x2 Dual-link), HDTV, D-sub, VIVO and Avivo."
http://www.diamondmm.com/p091206.php
Naively, that would put RV570 at 260M transistors.
Hmm, according to this:
http://www.techreport.com/onearticle.x/10789
RV570 is 80nm. I've honestly lost track of what previous rumours have suggested.
That would obviously increase transistor count somewhat: 330M transistors? But do we care, huh?
Jawed
Anyone care to guess why ATi is using 2 bridge connectors?
Is it only to ensure full speed AA? Triple CrossFire sounds a bit dodgy (SFR/AFR with three cores?) and CrossFire physics doesn't need a physical connection between the regular GPU and physics GPU. Or could it be something else, like making CrossFire cross-vendor compatible so it will work on every available chipset around that supports two physical 16x PCIe slots?
I read somewhere that ATi's version of the bridge was a 2-way bridge. So could it be that on SLI chipsets or (future) Intel chipsets GPU1 sends all the data through bridge1 to GPU2 which then renders his part and sends the data back through bridge2 to GPU1 which merges it into 1 image? That way no data would have to pass through the second PCIe 16x slot to the vidcard2 and the GPUs wouldn't need the PCIe bus to communicate with eachother. Or is that just wishful thinking? :P
Btw, the new X1900GTs have already made an appearance. On Dutch forum Gathering of Tweakers one poster has just posted pics of his new X1900GT which sports the new cooler and also shows CrossFire connectortraces on the PCB:
http://img97.imageshack.us/my.php?image=dsc01023kn6.jpg
http://img125.imageshack.us/my.php?image=dsc01021wo5.jpg
Source: http://gathering.tweakers.net/forum/list_message/26493019#26493019
chavvdarrr
15-Sep-2006, 20:14
Nearly. X1300 XT is 12/4/4! ;)
and what about X1650XT (or is it x1700 now?!)
Anon Lamer
15-Sep-2006, 22:08
"Diamond Multimedia (www.diamondmm.com), a leading manufacturer of PC graphics cards, sound cards and communications products, will begin shipping three new versions of its popular Viper video card line on September 14, including the X1950 CrossFire™ & X1950XTX in PCI-E models along with the X1950PRO available in an AGP version."
"Diamond Ships Six New Viper Cards
The X1950AGP model has a 256MB GDDR4 interface, core clock speed of 600MHz and a 1.4 GHz memory speed (Due Oct 2006). All these products support Dual DVI (x2 Dual-link), HDTV, D-sub, VIVO and Avivo."
http://www.diamondmm.com/p091206.php
GDDR4 on the bottom X1950 model? Is it already mainstream?
and what about X1650XT (or is it x1700 now?!)
24/8/8 IIRC.
X1900GT 36/16/12 Since when were there 16 TMUs? I've never seen that suggested before :huh:
trumphsiao
16-Sep-2006, 05:31
What Hkepc stated that RV570 have same Die Size of R580 is virtually wrong.
less than 65% R580 Die Size for sure.
vertex_shader
16-Sep-2006, 09:08
GDDR4 on the bottom X1950 model? Is it already mainstream?
They removed the x1950pro agp part.
Last year diamond already made a mistype with the x1800AGP :roll:
trumphsiao
20-Sep-2006, 11:36
sorry to tell you guys. Die Size of RV560 is bigger than that of G71.
Dunno why, maybe RV560 is cripple RV570.
Chinese Article did state RV570 is almost as big as R580.
R580 is nearly square. RV570 from that piccie is clearly narrower across the top than it is down the side.
sorry to tell you guys. Die Size of RV560 is bigger than that of G71.
Dunno why, maybe RV560 is cripple RV570.
Maybe RV570 is 80nm and RV560 is 90nm?
Maybe RV570 is 80nm and RV560 is 90nm?
Both are on 80nm, die shots are here (http://www.hkepc.com/bbs/itnews.php?tid=669674&starttime=0&endtime=0).
I wouldnt be suprised if RV560 is larger than G71 considering an RV535 (RV530 on 80nm) is 130mm² (http://www.elitebastards.com/cms/index.php?option=com_content&task=view&id=107&Itemid=27&limit=1&limitstart=1).
trumphsiao
22-Sep-2006, 09:01
Maybe RV570 is 80nm and RV560 is 90nm?
both of them are 36ALU DX9 chips.except Clock speed and MC difference.
RV560 is supposed to be 8-1-3-x, i.e. 24 ALU pipelines.
RV570 is supposed to be 12-1-3-1, i.e. 36 ALU pipelines.
Jawed
trumphsiao
22-Sep-2006, 13:00
RV560 is supposed to be 8-1-3-x, i.e. 24 ALU pipelines.
RV570 is supposed to be 12-1-3-1, i.e. 36 ALU pipelines.
Jawed
dunno why RV560 have bigger than G71 die size.:sad:
dunno why RV560 have bigger than G71 die size.:sad:
Hmmm. Let me do a little back of the envelope scratching...
Interpolating between R580 and RV530 we have:
Size (mm2) Chip
150 RV530 on 90nm
352 R580 on 90nm
196 G71 on 90nm
217 RV560 on 90nm interpolation
285 RV570 on 90nm interpolation
172 RV560 on 80nm?
225 RV570 on 80nm?
So it does seem RV560 should be smaller than G71 based on an interpolation using only the number of pixel shader units. So I am going to guess that there is more in there, such as a compositing circuit.
Please anyone feel free to correct my errors/wrong assumptions.
ERK
dunno why RV560 have bigger than G71 die size.:sad:
My guess, probably, it might concern the ring-bus programmable memory controller on the RV560. If it happens to be the same controller as R520, it would take a big area on the core (as I remembered that the MC of R520 takes quite big area on core too, please anyone correct me if it is wrong).
RV530/535 has the same MC, just 128bit (so, half the bus width). RV560 is also supposed to be 128bit. RV570 is supposed to be 256bit.
ERK, you might want to add the 130mm^2 80nm RV535. Then it's simpler to think of RV560 as a little under double that (MC "width" isn't doubled).
Wow, taking the 130 mm2 for RV535 and doubling to 260 mm2 is a big difference from my 172 mm2 interpolation. I'm getting the feeling that this whole excercise may be pointless. :)
Well, obviously things don't scale linearly--not everything, anyway. If those RV570 die shots are right, and it's 67% of R580, then it's about 235mm². So I'm not sure how RV560, with 1/3 fewer fragment shaders, will be anywhere near 260mm².
Yeah, I tried those calcs last night and just gave up. :) Too many variables I don't know about. But just looking at RV535/RV530 (130mm²/150mm²=~87%, pretty close to 80/90nm=~89%), you're a bit generous with the 90-80nm transition.
R520 288mm² vs. R580 352mm² might help, but I don't know how. All else equal except the fragment shader units, you can extrapolate a *very* rough 2mm² per fragment shader unit, but that figure doesn't get us anywhere near RV570's apparent die size. 352mm²-16*2mm²=322mm². 322mm²*80nm/90nm=~285mm². Even using RV535/RV530 rather than 80nm/90nm, we only reach 279mm². That's still a long way off the estimated .67% of 352mm², or 236mm², but I didn't account for omitting some vertex shaders and perhaps cache (RV570 probably being optimized for a lower res than R580).
trumphsiao
23-Sep-2006, 00:35
I measure up the Die Size of RV560 is roughly 125% that of G71
Blazkowicz
24-Sep-2006, 12:39
Yeah, I tried those calcs last night and just gave up. :) Too many variables I don't know about. But just looking at RV535/RV530 (130mm²/150mm²=~87%, pretty close to 80/90nm=~89%), you're a bit generous with the 90-80nm transition.
nitpick : shouldn't it be (80/90)² ? (which is 79%)
Prrrobably. :oops: That makes RV530->RV535 all the more puzzling. It does get us closer to RV570's estimated die size, though. RV560 remains inscrutable.
vertex_shader
03-Oct-2006, 13:53
Looks like the launch delayed to october 30. :roll:
http://www.ocwforums.com/bbs/showthread.php?threadid=55139
Looks like the launch delayed to october 30. :roll:
http://www.ocwforums.com/bbs/showthread.php?threadid=55139
According to the source, the X1950Pro is still scheduled for the 17th of October. The X1650XT has been confirmed to be delayed until the 30th of October because of 'technical issues'. According to ATI it should be a hard launch, but rumors have it that this could be a small problem.
vertex_shader
06-Oct-2006, 12:18
According to the source, the X1950Pro is still scheduled for the 17th of October. The X1650XT has been confirmed to be delayed until the 30th of October because of 'technical issues'. According to ATI it should be a hard launch, but rumors have it that this could be a small problem.
Now ocworkbench say the same :lol:
"A few days back we reported that the launch is postponed to 30 Oct. I received news from ATi that the two cards were planned to be launched on seperate dates. ATi X1950Pro on 17 Oct and X1650XT on 30 Oct."
http://www.ocwforums.com/bbs/showthread.php?threadid=55262
Club 3D Radeon X1650XT pictured:
http://img10.picsplace.to/17/thumbs/x1650xt.jpg (http://img11.picsplace.to/img10/17/x1650xt.jpg)
probably reference design PCB
trumphsiao
06-Oct-2006, 19:40
Club 3D Radeon X1650XT pictured:
http://img10.picsplace.to/17/thumbs/x1650xt.jpg (http://img11.picsplace.to/img10/17/x1650xt.jpg)
probably reference design PCB
I dunno to which extent of possibility drove of people would buy RV560 instead of 7900GS on same price segment ?
I dunno to which extent of possibility drove of people would buy RV560 instead of 7900GS on same price segment ?
If X1650XT is available at the suggested €150 price segment while the 7900GS is selling around €199, then the X1650XT is a pretty attractive card at that pricepoint which is faster than the GF7600GT.
Several shops have already listed the X1950Pro at about €199, so that's the direct competitor to the 7900GS. And you can bet on it that the X1650XT will be prices a bit lower than that.
I dunno to which extent of possibility drove of people would buy RV560 instead of 7900GS on same price segment ?
Based on Finnish prices, I doubt RV560 would end up in the same price segment, ATM in the same price segment with 7900GS is X1900GT, which trashes it, and in future, there will be X1950Pro which again, trashes the 7900GS.
trumphsiao
08-Oct-2006, 11:17
Based on Finnish prices, I doubt RV560 would end up in the same price segment, ATM in the same price segment with 7900GS is X1900GT, which trashes it, and in future, there will be X1950Pro which again, trashes the 7900GS.
just got prompt on X1900GT price in China Retail Channel . wow,It's a none-brainer deal (1400RMB=USD 180 ) hope this time ATI would discover prepatent solution to solve RV560 less-sufficient TMU Problem.
http://www.dailytech.com/article.aspx?newsid=4472
Brief overview with benchmarks.
Jawed
It's a tad disappointing to see it without dual-link DVI on at least one port, if retail examples will follow the DT one (and indeed they're correct in what they say). I know I'm not the norm, but I'm always on the lookout for a cheap, v.quiet, somewhat frugal mid-range board, but with the caveat that it can drive my big-ass Dell.
Very strange if the dual-link DVI has disappeared.
It's not a side-effect of the new CrossFire connectivity is it?...
Jawed
There should be no output limitations with Crossfire now, present or future. Either they're wrong, or I'd guess it's a pre-production sample curiosity at most.
trumphsiao
09-Oct-2006, 13:21
There should be no output limitations with Crossfire now, present or future. Either they're wrong, or I'd guess it's a pre-production sample curiosity at most.
I wager this time RV570 Crossfire efficacious performance shall be better than
X1800XT solution . But It's too late sans any better Chance to encounter Price-Cut 7900GS/7950GT
Now doubt about it, it's late - several months. Once G80 is on the market (2 weeks after RV570), RV570 will be competing against G71/7900GTX/GTO.
Makes me wonder if the lateness of this has been impacted by trying to make R600 less late.
Jawed
*Update* The Radeon X1950 Pro has internal dual-link TMDS transmitters for both DVI outputs. HDCP is also supported on the reference board.
:smile:
Ditching dual-link DVI should have been ridiculous on its face, as I thought it was a part of AVIVO. Yeah, I thought if they did ditch it, it would be a byproduct of internalizing XFire. Much ado.
I have a hard time believing RV570 is as fast as R580, though, without seeing the 256MB as a hugely limiting factor. They should've tried 19x12 or something, and with 1x or 2x AA, to minimise potential memory limitations, but this /is/ just a preview.
The drop in power draw is awesome, though. Surprisingly high, actually.
Edit: Yeah, geo, I meant a surprisingly high drop, not surprisingly high draw. 54W seems crazy large for just a die change. Then again, the fact that it seems crazy just highlights (http://www.xbitlabs.com/articles/video/display/power-noise_4.html) my ignorance (http://www.techreport.com/reviews/2006q3/geforce-7950gt/index.x?pg=9), as apparently the current ("quad"-disabled R580) GT already draws 40W less than the XT 256. That puts any savings from the process drop in the 10-15W range, if DT's overall system draw #s are comparable to TR's. Not bad, but not monumental.
And there's something wrong with DT's #s, as TR shows (http://www.techreport.com/reviews/2006q3/geforce-7950gt/index.x?pg=8) the XT 256 to hit 5.7k in 3DM06, not a mere 5k (which is closer to the current GT's score). If they're using review drivers, maybe those are (severely!) impairing the XT 256.
The drop in power draw is awesome, though. Surprisingly high, actually.
Those power draw numbers they showed appeared to only make sense as system power draw rather than just vid card draw, even tho not labelled that way. If that is correct, then a 54W drop off just based on the vid card is a yowza kind of number, as from a percentage pov a good bit of the overall system number is static.
Ditching dual-link DVI should have been ridiculous on its face, as I thought it was a part of AVIVO. Yeah, I thought if they did ditch it, it would be a byproduct of internalizing XFire. Much ado.
I have a hard time believing RV570 is as fast as R580, though, without seeing the 256MB as a hugely limiting factor. They should've tried 19x12 or something, and with 1x or 2x AA, to minimise potential memory limitations, but this /is/ just a preview.
The drop in power draw is awesome, though. Surprisingly high, actually.
Edit: Yeah, geo, I meant a surprisingly high drop, not surprisingly high draw. 54W seems crazy large for just a die change. Then again, the fact that it seems crazy just highlights (http://www.xbitlabs.com/articles/video/display/power-noise_4.html) my ignorance (http://www.techreport.com/reviews/2006q3/geforce-7950gt/index.x?pg=9), as apparently the current ("quad"-disabled R580) GT already draws 40W less than the XT 256. That puts any savings from the process drop in the 10-15W range, if DT's overall system draw #s are comparable to TR's. Not bad, but not monumental.
And there's something wrong with DT's #s, as TR shows (http://www.techreport.com/reviews/2006q3/geforce-7950gt/index.x?pg=8) the XT 256 to hit 5.7k in 3DM06, not a mere 5k (which is closer to the current GT's score). If they're using review drivers, maybe those are (severely!) impairing the XT 256.
maybe the performance is because RV570 is 12-1-3-2? :o
vertex_shader
11-Oct-2006, 13:00
http://img96.imageshack.us/img96/8813/asusx1950proai4.jpg
Arnold Beckenbauer
11-Oct-2006, 13:39
http://www.digit-life.com/articles2/digest3d/index0609.html
Check the results. :twisted:
INKster
11-Oct-2006, 13:47
http://www.digit-life.com/articles2/digest3d/index0609.html
Check the results. :twisted:
There's no RV570/RV560 (X1950 Pro/X1650 XT) in there...
Arnold Beckenbauer
11-Oct-2006, 13:56
There's no RV570/RV560 (X1950 Pro/X1650 XT) in there...
Oh really?
http://www.digit-life.com/articles2/digest3d/0906/itogi-video-m62-wxp-pcie-aaa.html
:wink:
vertex_shader
11-Oct-2006, 14:14
Oh really?
http://www.digit-life.com/articles2/digest3d/0906/itogi-video-m62-wxp-pcie-aaa.html
:wink:
3dmark2k6 sm3.0 1280x1024 4x aa 16x af x1900gt rev1 (575/600) 855 point, x1950pro (580/1400)1365 point, great boost :)
The RV560 numbers are nothing to sniff at, either! 805 points! For only 22.4GB/s bandwidth that's pretty nice (as compared to X1900GT's 38GB/s).
Jawed
I have a hard time believing RV570 is as fast as R580, though, without seeing the 256MB as a hugely limiting factor. They should've tried 19x12 or something, and with 1x or 2x AA, to minimise potential memory limitations, but this /is/ just a preview.
It's interesting. From the above Digit-Life link:
3DM06 1600x1200 4xAA 16xAF
X1900XT 256MB 625/1450 - 1105
X1950Pro 256MB 580/1400 - 985
3DM06 1600x1200
X1900XT 256MB 625/1450 - 1815
x1950Pro 256MB 580/1400 - 1610
Roughly in line with clock speeds.
Arnold Beckenbauer
11-Oct-2006, 14:41
There are results of Far Cry, Doom 3, FEAR etc, not only 3D2006.
Far Cry with AA/AF for example: http://www.digit-life.com/articles2/digest3d/0906/itogi-video-fc1-wxp-pcie-aaa.html
X1950Pro is a nice competitor for 7900GS.
INKster
11-Oct-2006, 14:47
There are results of Far Cry, Doom 3, FEAR etc, not only 3D2006.
Far Cry with AA/AF for example: http://www.digit-life.com/articles2/digest3d/0906/itogi-video-fc1-wxp-pcie-aaa.html
X1950Pro is a nice competitor for 7900GS.
Maybe for 7900 GS, yes.
But what it NV decides to kill that SKU, and lower the 7950 GT down to that price-point ?
G73-B1 is coming, and since it's 80nm also, they could crank the speed of that part to compensate for the 7900 GS, and let 7950 GT take its place.
It's not like it can't be done, look at the 7900 GTO vs GTX. It even manages to maintain the same core clock and cooling solution as the high-end.
In that sense, is there any room for OC in those RV570's with the standard cooler ?
vertex_shader
11-Oct-2006, 15:05
There are results of Far Cry, Doom 3, FEAR etc, not only 3D2006.
Far Cry with AA/AF for example: http://www.digit-life.com/articles2/digest3d/0906/itogi-video-fc1-wxp-pcie-aaa.html
X1950Pro is a nice competitor for 7900GS.
Some german shop already pre listed the asus x1950pro 256mb around 168-170 euro http://www.xada.de/shop2/index3.php?xlink=35955&ref=8 http://www.fortknox.de/cat3.php?art=85955.
7900gs price now around 175-185 euro.
Yeah, I don't get why RV570 is so close to the XT 256 in 3DM06 or CoD2, both with and without AA. I mean, CoD2 can't be bandwidth-limited, b/c even the X1950XT barely distances itself from the X1900XT. It's not ROPs or TMUs and probably not pixel shaders (see X1800XT). Maybe a vertex limitation? Must be the demo itself. As for 3DM06, I can't figure out how RV570 distances itself so handily from the X1900 GT, and why it's only 10% shy of the XT 256. Sure, the XT's hobbled by RAM, but is that really it? It looks like the difference is more in line with oure core clocks than anything else. Otherwise, RV570's spotting R580 33% more ROPs, texture units, and pixel shaders. Dunno about vertex shaders, but I thought '05 was setup limited, not '06. It's veird!
Maybe the A64 is a limiting factor, but DT was using a C2D when they got 5k out of their XT. Xbit Labs got 5k from an XT 256 but with an A64. TR and now FS (http://www.firingsquad.com/hardware/evga_e-geforce_7900_gto/page4.asp) both pulled 5700 with the same C2D with which DT manages just 5k. Friggin' lack of standards. :)
Aaanyway, the XT more properly (well, at least more in line with the theoretical differences) distances itself in everything else D-L benches, even 3DM05. Something just wacky with 3DM06 and CoD2, I guess, and probably more to do with the drivers in the first case and the demo (.exe) in the second.
Edit: Welcome to the forums, Arnold, and thanks for the link.
Arnold Beckenbauer
11-Oct-2006, 23:07
Some german shop already pre listed the asus x1950pro 256mb around 168-170 euro http://www.xada.de/shop2/index3.php?xlink=35955&ref=8 http://www.fortknox.de/cat3.php?art=85955.
7900gs price now around 175-185 euro.
The X1950Pro is listed in some shops since September 23rd; Xada and Fortknox belong together. Such offers are normal: pre-order, pay and wait. You will get your X1950Pro, when it coasts, what you have paid, and it doesn't matter, that X1950Pro is available.
trumphsiao
12-Oct-2006, 15:26
The X1950Pro is listed in some shops since September 23rd; Xada and Fortknox belong together. Such offers are normal: pre-order, pay and wait. You will get your X1950Pro, when it coasts, what you have paid, and it doesn't matter, that X1950Pro is available.
http://we.pcinlife.com/attachments/forumid_206/Snap2_jkjj1hb9TD7O.jpg
:roll:
Arnold Beckenbauer
12-Oct-2006, 15:36
I know, that the RV5x0 graphics cards are not available until October 17th (or later). What I have meant, is that is senseless to pre-order X1950Pro in these german shops.
Example: You pre-order a X1950Pro today and pay 190,- Euro. Next tuesday the new ATi video cards are available in these shops, but they cost 200 € - 10 Euro more. Ergo: You will get your pre-ordered X1950Pro, when it costs 190,-.
trumphsiao
13-Oct-2006, 20:19
I know, that the RV5x0 graphics cards are not available until October 17th (or later). What I have meant, is that is senseless to pre-order X1950Pro in these german shops.
Example: You pre-order a X1950Pro today and pay 190,- Euro. Next tuesday the new ATi video cards are available in these shops, but they cost 200 € - 10 Euro more. Ergo: You will get your pre-ordered X1950Pro, when it costs 190,-.
Die Size of RV570 is less than 238mm^2
HIS X1950 PRO ICEQ3 'Turbo' 256 MB- 620/1480 - VIVO HDCP 2X DVI ?
http://www.us.ncix.com/products/index.php?sku=21014&vpn=H195PRQT256DV-R&manufacture=HIS&nocookie=1&membership=0
bye
http://www.dailytech.com/article.aspx?newsid=4666
575MHz core
675MHz 128-bit memory (a bit slower than RV530, wierd)
Incorrectly states 12 pipelines though, should be 8 if it follows 8-1-3-1/2.
Jawed
INKster
24-Oct-2006, 00:27
I have my doubts that this will be enough to counter a highly clocked G73-B1 (80nm, shipping now as the "Geforce Go 7700" on a few Asus Laptops).
Let's wait and see.
Yeah, that 24ps to 12 ROPs clanged off my ear as well.
I have my doubts that this will be enough to counter a highly clocked G73-B1 (80nm, shipping now as the "Geforce Go 7700" on a few Asus Laptops).
Let's wait and see.
I have my doubts that we will see Geforce 7700 on the desktop side of things. :wink:
INKster
24-Oct-2006, 02:00
I have my doubts that we will see Geforce 7700 on the desktop side of things. :wink:
Depends of the G84 timetable.
If it comes out in H1'07, then the G73-B1 is redundant.
But..., i know of a certain desktop card already equiped with a G73-B1 GPU.
Only that right now, it's being marketed as a "regular" 7600 GT. :D
http://64.233.179.104/translate_c?hl=pt-PT&ie=UTF-8&oe=UTF-8&langpair=zh-CN%7Cen&u=http://www.pconline.com.cn/diy/factory/0610/888218.html&prev=/language_tools
vBulletin® v3.8.6, Copyright ©2000-2013, Jelsoft Enterprises Ltd.