The Official RV630/RV610 Rumours & Speculation Thread

Status
Not open for further replies.
1 - Why the PCB of the HD2600XT GDDR4 is more complex than the PCB of the HD2600XT GDDR3?

2 - Why the cooler of the HD2600XT GDDR4 is more powerful than of the HD2600XT GDDR3? And why same core clocks?

3 - Why the HD2400XT has hardware Crossfire and the HD2600PRO haven't it? What the sense of this?

I'm waiting that the HD 2600XT GDDR3 costs less than the GF 8600GT. At least some models (like Sapphire 4-layer PCB).
And the only dual-GPU that I liked was Gecube's.

bye
 
Any idea what one might expect from an RX2600PRO-T2D256Z (AGP) relative to a 8600. At €200 this looks like a nice upgrade.
http://www.heise.de/computex/bilderstrecke/99/4
Sorry if mentioning AGP has given you an aneurysm.

Uhh thats overpriced, 200euro too much for this card, hd2600pro coming with ddr2memory (some AIB will use gddr3 memory on hd2600pro, but for 200euro this cards never has any chance to sell), 1950gt agp will kill this card and much cheaper, can't do dx10 but who cares now?
HD2xxx cards sound and HDTV features not really a user magnet, when from 100 user 5 care about this in first place.

Performance rumors not suggest any good for AMD, months later and the gddr4 version still can't beat or slightly can beat the 8600gts, who cares when its has a better price/performance ratio than the 8600gts when 8600gts already has bad performance in some games at 1280x1024 resoluton, and most of the users who buy midrange cards has tft monitors with this native resolution.
I think nv and amd playing a bad joke with users right now, they are try sell low performance in midrange price range with the magic word "directx10" :cry:

Here is my quess what this cards worth (we se no real reviews yet, i place my quess for performance rumors and some early mostly 3dmark test, so take this with a grain of salt): hd2400pro-xt 256mb (64bit,ddr2) ~60-65euro, hd2600pro (128bit,ddr2) 256mb 85euro, hd2600xt (128bit,gddr3) 256-512mb 115-125euro, hd2600xt (128bit,gddr4) 256-512mb 135-155 euro.
Some AIB's will make cards whats not following the reference, like HIS hd2600pro with gddr3 memory, and powercolor "worst joke ever card" the hd2400xt 64bit with gddr3 memory.

X1950gt/X1950pro still the best performance/price deal now in this range, and this not change in the summer, maybe we see some real midrange performance card for the back to school season, hopefully AMD can release a performance cards in time, and not miss the money bag again ;)
 
Last edited by a moderator:
Looks like Powercolor using custom cooler too in the hd2600xt gddr4 version, and not the single slot reference cooler
2.jpg


hd2600pro
4.jpg
 
ELSA cards
elsa_240XT.jpg

elsa_260pro.jpg

elsa_260xt.jpg


Jetway hd2600xt gddr4
jetway_x26xt.jpg


Some more powercolor card
hd2400pro
powercolor_2400scs.jpg


hd2600pro agp
powercolor_2600proagp.jpg

Link




Looks like most of AMD AIB's using they own PCB designs and custom coolers.
 
2600XT GDDR3 Preview @ PConline

Google-translation

I'm almost speechless :cry:

In nfs carbon its can match up the 8600gts, hl2 little slower than 8600gt, in the other games much slower than the 8600gt, this is a disaster.

This performance will be good for a hd2400pro card, but a shame for a hd2600xt, and i don't think so gddr4 version change the picture.
AMD forget how they can make gamer cards, i'm deeply dissapointed, i not belive the diver wonder, i not except any good for rv670 anymore.

Well, hd2600xt can be a hotseller for 80euro-100$ :smile:
 
Which drivers are they testing on? No hint on these in that preview (there are no results for the DDR4 XT version, however).
Very strange results, anyway: RV630 is not so texture limited as R600 (relatively speaking) and with 800 MHz core speed... Even 3Dmark score is strangely very low compared to the worse rumors.
RBE limited? :-?
 
There aren't any photos, only two pictures of HD2600XT from ATi... any single screenshot from control panel... Maybe I'm too optimistic, but it lookes like PRO's results to me. RV630's shader core and even texturing engine are less crippled than G84's shader core and texturing engine (compared to R600 and G80) - results should be quite different. I can't get it.
 
There aren't any photos, only two pictures of HD2600XT from ATi... any single screenshot from control panel... Maybe I'm too optimistic, but it lookes like PRO's results to me. RV630's shader core and even texturing engine are less crippled than G84's shader core and texturing engine (compared to R600 and G80) - results should be quite different. I can't get it.

When i first read the test i think its mistake and the card is a hd2600pro ddr2 in the test, but everywhere they write xt, so if a mistake than a huge one, or its the translator engine fault :LOL:
 
More interestingly, apparently AMD sources are whispering to Fudo that they've scored 60% of the OEM business this time around. http://www.fudzilla.com/index.php?option=com_content&task=view&id=1360&Itemid=1 Which would go to the conversation upstream about back to school sku timing, TSMC capacity, and (possibly) why RV630/RV610 channel avail got pushed back some from the original expectation.
Yeah, well one hopes so for AMDs struggling bottom line.

Pity they seem to have surrendered a swathe of the notebook market again to NV, having lost the MacBook refresh and early Santa Rosa design wins. I still haven't seen any shipping 690-based notebooks and it's been a good 3 months since the announcement -- would have thought that the power savings and bundling options (Turion) would have made it an attractive proposition to OEMs?
 
vertex_shader: There isn't any photo or screenshot in this reveiw, which would confirm, that they really tested XT. And according to some sources, production of XTs was just initiated (XT should use the last chip revision). Only PROs are already in production. Or am I wrong? Shortly - even computex XT's were only samples and manufaturers said, that they are waiting for final revision of RV630. So whatever was tested in this review, it couldn't be HD2600XT with final silicon.
 
vertex_shader: There isn't any photo or screenshot in this reveiw, which would confirm, that they really tested XT. And according to some sources, production of XTs was just initiated (XT should use the last chip revision). Only PROs are already in production. Or am I wrong? Shortly - even computex XT's were only samples and manufaturers said, that they are waiting for final revision of RV630. So whatever was tested in this review, it couldn't be HD2600XT with final silicon.

The weird thing is they not mention anything about drivers, in the test configuration amd driver name missing.
Anyway when fud or not, till weeks left until performance NDA expire.
 
Next strange thing:

Hardspell: HD2600XT GDDR4 (800/1100): 3DM06 SM3.0 score: 2105
PCOnline: HD2600XT GDDR3 (800/700): 3DM06 SM3.0 score: 1627

3DMark06 isn't bandwidth limited (overclocking of memory makes only subtle difference, nice example is X1900XTX/X1950XTX difference, which is about 3%). Anyway, RV630 shouldn't be BW limited - it has only 4 ROPs and 8 TMUs...

difference between pconline and hardspell is simply too high (29%) to be caused only by faster memory modules. I still think, that the most likely scenario is that pconline tested HD2600PRO, which is 600/500. XT's core is clocked 33% faster (800), which corresponds to 29% difference in 3DM06.

(I consider hardspell's review as reliabe - they have photos, screenshots, CCC etc.)
 
Status
Not open for further replies.
Back
Top