View Full Version : Mainstream Video Quality Shootout
Alan finally got around to publish his latest article, Mainstream Video Quality Shootout: ATI vs NVIDIA & XGI (http://firingsquad.com/hardware/ati_nvidia_xgi_mainstream_video_quality_comparison/).
And for the lazy ones (;))
"Our sources indicate that ATI is testing new drivers that will use more of the video hardware in the X1800 and the X800. The buzz we’re getting from our sources is that this new driver may bring ATI’s video performance into the triple digits in the HQV Benchmark!"
:idea:
Y'know, I gave Alan's video work "specialty niche reporting of the year" for 2005, and it certainly was deserving. . .but it would be so much more satisfying if this piece was 7600 vs X1600.
I would have gladly waited another month, or two, for that. :sigh:
Skrying
08-Dec-2005, 00:27
Interesting, but the products used dont really make sense to me. I think a much better comparison would be 7800GT(X) X1800XL(XT) and the XGI product of your choice (dont know the lineup). Since the X800XL does not represent the most current video quality solution from ATi, and apparently the 6600GT does not either for Nvidia.
The word "mainstream" is your clue there --he was after a pricing thing for the comparison. Hence my sigh --we ought to be just about to see new mainstream parts (NV) or they are just becoming available (ATI).
Skrying
08-Dec-2005, 00:36
Then was it truely impossible to find an X1300? The range doesnt make sense to me, no matter how you slice it.
It was still a good read, and I have a 9600 in my htpc :( Ah well, I am somewhat dissapointed about the 6600 thing b/c my 6200 is of course unlocked and was sdestined for being the HTPC card later, but unfortunately the drivers won't allow pure video to function on a 6200 even if it is essentially a 6600 ah well...
Then was it truely impossible to find an X1300? The range doesnt make sense to me, no matter how you slice it.
Yes, well, X1300 is "low-end". Now, before you tell me how can the XGI card *not* be low-end at that price, I shall quickly exit stage left with my fingers in my ears going "la la la, I can't hear you!" :razz:
Skrying
08-Dec-2005, 00:51
It was still a good read, and I have a 9600 in my htpc :( Ah well, I am somewhat dissapointed about the 6600 thing b/c my 6200 is of course unlocked and was sdestined for being the HTPC card later, but unfortunately the drivers won't allow pure video to function on a 6200 even if it is essentially a 6600 ah well...
You sure? My 6800 is currently away at the BFG having something down or fixed or looked at or whatever they do on the RMA process, and all I have in my rig right now is a MX440 (and my other rig is torn down for upgrade time!) but I've always been confused about what is needed for pure video. I know there is a driver for it, or something like that, but isnt there another piece of not so free software that is required?
You sure? My 6800 is currently away at the BFG having something down or fixed or looked at or whatever they do on the RMA process, and all I have in my rig right now is a MX440 (and my other rig is torn down for upgrade time!) but I've always been confused about what is needed for pure video. I know there is a driver for it, or something like that, but isnt there another piece of not so free software that is required?
I believe they charge $10 for the decoder. That's my memory anyway. I also seem to recall there are Issues with 6800 and pure video.
Edit: Urrp! $19.95 for the cheapie, and up to $49.95 if you want some Dolby goodness. Tho there is a free 30 day trial if you want to see if you like it. And it was the GT/Ultra that have decode problems. They aren't accelerated for decode.
Skrying
08-Dec-2005, 01:01
Thank's Geo, that seems to be what I recall.
I thought it was $20.00, but it doesn't really matter.
On newer cards it uses the hardware to do stuff with the video, and they said that it will only function on a minimum of a 6600 due to the 6200 not having enough power. I got my 6200 with the idea of putting it in my htpc later and did not know that at the time it was a dissapointment especially since as I said it is unlocked, of course it was only $50.00 so that is still a cheap card and I cannot complain too much.
Here (http://www.nvidia.com/page/purevideo_support.html) is Nvidias info, if you scroll down you will see what I mean, the 6200 pci-e does do some of the stuff, but it misses out on other bits that the rest of 6 series does. Basically I don't want to buy it (purevideo) and then see if it will work, I would rather know in advance...
It seems to me that ATI has been resting on its laurels for years now. Pretty sad.
Jawed
Skrying
08-Dec-2005, 17:19
It seems to me that ATI has been resting on its laurels for years now. Pretty sad.
Jawed
AVIVO. The ATi card tested did not have it, so its not accurate representation of ATi's current video quality output.
No, but with talk of improved video decoding coming to non-AVIVO cards, it seems to me that ATI has waited to get called-out on poor quality decoding before acting.
Jawed
mrcorbo
09-Dec-2005, 00:51
I thought it was $20.00, but it doesn't really matter.
On newer cards it uses the hardware to do stuff with the video, and they said that it will only function on a minimum of a 6600 due to the 6200 not having enough power. I got my 6200 with the idea of putting it in my htpc later and did not know that at the time it was a dissapointment especially since as I said it is unlocked, of course it was only $50.00 so that is still a cheap card and I cannot complain too much.
Here (http://www.nvidia.com/page/purevideo_support.html) is Nvidias info, if you scroll down you will see what I mean, the 6200 pci-e does do some of the stuff, but it misses out on other bits that the rest of 6 series does. Basically I don't want to buy it (purevideo) and then see if it will work, I would rather know in advance...
Here's the killer about the page you referenced. It's wrong. See the PCI-e table? The capabilities of the 6800Ulta, 6800GT and 6800 vanilla are all incorrect as far as their support for WMV acceleration. The 6800U/GT PCIe are based on the NV40 chip like the AGP versions and so have the same broken VPP. The 6800 PCI-e is based on a different chip and actually has a fully functional VPP, so is MORE capable than the higher-end versions. Lots of heat on the AVS forum about this! All the AGP 6800s have a broken VPP and all the 6600s PCIe and AGP have a working one.
All the AGP 6800s have a broken VPP and all the 6600s PCIe and AGP have a working one.
Yeah but the pci-e 6200 has a working one as well (at least the first version that was the same chip as the 6600) and that is what is kiling me :), I don't care about the other bit b/c it doesn't affect me. But the 6800GS should probably work as well.
Murakami
09-Dec-2005, 21:02
Here's the killer about the page you referenced. It's wrong. See the PCI-e table? The capabilities of the 6800Ulta, 6800GT and 6800 vanilla are all incorrect as far as their support for WMV acceleration. The 6800U/GT PCIe are based on the NV40 chip like the AGP versions and so have the same broken VPP. The 6800 PCI-e is based on a different chip and actually has a fully functional VPP, so is MORE capable than the higher-end versions. Lots of heat on the AVS forum about this! All the AGP 6800s have a broken VPP and all the 6600s PCIe and AGP have a working one.
Can you link something about that? I'm very interested (i own a 6800 GT PCI-E), i cannot believe that nVidia keep on lying...
Y'know, I gave Alan's video work "specialty niche reporting of the year" for 2005, and it certainly was deserving. . .but it would be so much more satisfying if this piece was 7600 vs X1600.
I would have gladly waited another month, or two, for that. :sigh:
I think when ATI releases those promised drivers, there will be a re-match. And if we're lucky we might have a 7600 by then. Wishing on both accounts ofcourse. :smile:
Murakami
10-Dec-2005, 13:42
Can you link something about that? I'm very interested (i own a 6800 GT PCI-E), i cannot believe that nVidia keep on lying...
No one? :cry:
Deathlike2
10-Dec-2005, 13:56
No one?
That's what a quick Google can do for you...
http://techreport.com/onearticle.x/7535
http://www.anandtech.com/video/showdoc.aspx?i=2305&p=11
http://www.overclockers.com/tips00681/
Remember that this (hardware flaw/bug) only applies to the NV40 (GeForce 6800GT/Ultra AGP) and NV45 (GeForce 6800GT/Ultra PCIe), NV41 (GeForce 6800) has fully function WMV9 decode acceleration.
Stuff in bold is what I added to what "this" is referencing.
NVidia didn't "lie".. but from reading their official response (and reading between the lines), you can extropolate that there is a hardware flaw. Like any company that has a flaw in their product.. it isn't "said" until a "later date".
I think when ATI releases those promised drivers, there will be a re-match. And if we're lucky we might have a 7600 by then. Wishing on both accounts ofcourse. :smile:
It's beginning to look a lot like Christmas, everywhere you go. . . err, that is, I'm now expecting "those drivers" before too long. And not expecting to see 7600 until January. But early Feb for the rematch would work for me. Tho I think Alan is so genuinely excited by all this stuff (which I always love to see) that I wouldn't be overly surprised if he gives us at least a little peak at what he's found as soon as he gets his mitts on the new drivers.
Murakami
10-Dec-2005, 18:02
That's what a quick Google can do for you...
http://techreport.com/onearticle.x/7535
http://www.anandtech.com/video/showdoc.aspx?i=2305&p=11
http://www.overclockers.com/tips00681/
Stuff in bold is what I added to what "this" is referencing.
NVidia didn't "lie".. but from reading their official response (and reading between the lines), you can extropolate that there is a hardware flaw. Like any company that has a flaw in their product.. it isn't "said" until a "later date".
I've already read those reviews (only Anand's one is a real review...:roll: ), they're too much old; if nVidia didn't fix this in driver i don't understand why their support page lists NV40 AGP as "broken" and his PCI-E counterpart (NV45) as "fully functional"...it's a non-sense (let's list as "broken" or "fully functional" both! :lol: ).
However, if the "broken part" it's only the WMV9 acceleration i can tolerate this one, cause the image quality is the same as newer chips.
Deathlike2
10-Dec-2005, 19:29
I've already read those reviews (only Anand's one is a real review... ), they're too much old; if nVidia didn't fix this in driver i don't understand why their support page lists NV40 AGP as "broken" and his PCI-E counterpart (NV45) as "fully functional"...it's a non-sense (let's list as "broken" or "fully functional" both! ).
However, if the "broken part" it's only the WMV9 acceleration i can tolerate this one, cause the image quality is the same as newer chips.
Well, this info is old because this issue was already revealed "that much time ago"... this is not driver fixable.. that's for sure. I don't like Anandtech for information.. but this info was pretty available for all major websites.. so I don't see the point in your response.
The G70, NV41 (6600 AGP), NV43 (6600 PCI-E), 6800 vanilla, and NV42 (6800GS) is not affected by this issue (the AGP/PCI-E version has no significance in the matter). The issue is just over the CPU utilzation (since it cannot be done on the video card, it will be done over the CPU)... there should be no image quality differences in the first place...
http://forums.nvidia.com/index.php?showtopic=530&st=0
If you're not into looking into Anandtech's info.. there's always NVidia's forums...
Murakami
10-Dec-2005, 20:13
Well, this info is old because this issue was already revealed "that much time ago"... this is not driver fixable.. that's for sure. I don't like Anandtech for information.. but this info was pretty available for all major websites.. so I don't see the point in your response.
The G70, NV41 (6600 AGP), NV43 (6600 PCI-E), 6800 vanilla, and NV42 (6800GS) is not affected by this issue (the AGP/PCI-E version has no significance in the matter). The issue is just over the CPU utilzation (since it cannot be done on the video card, it will be done over the CPU)... there should be no image quality differences in the first place...
http://forums.nvidia.com/index.php?showtopic=530&st=0
If you're not into looking into Anandtech's info.. there's always NVidia's forums...
Your post does not explain why nVidia lists, in his PureVideo support page, NV40 as "broken" and NV45 as "fully functional": please, explain me that thing, not NV41, 42, 43 and so on...if you're right, nVidia keeps on lying, no excuse or reading between the lines.
I would like to see new tests made with new drivers, new PureVideo decoder and 6600 GT PCI-E versus 6800 GT PCI-E (not AGP)...:evil:
Deathlike2
10-Dec-2005, 20:37
Your post does not explain why nVidia lists, in his PureVideo support page, NV40 as "broken" and NV45 as "fully functional": please, explain me that thing, not NV41, 42, 43 and so on...if you're right, nVidia keep on lying, no excuse or reading between the lines.
I would like to see new tests made with new drivers, new PureVideo decoder and 6600 GT PCI-E versus 6800 GT PCI-E (not AGP)...
I didn't see the link... apparently NVidia made a mistake in their listing..
NV40 (Geforce 6800 Ultra/GT AGP) and NV45 (Geforce 6800 Ultra/GT PCI-E) are the hardware with the bug... I'll cut them a "little slack" as their tables are huge.. though it should be their job to get the info correct.
Did I not say that new drivers won't fix the FLAW in the hardware? You will not get any performance benefits if the hardware is not able to support it...
Skrying
10-Dec-2005, 20:41
Nv40 covers 6800 Ultra, 6800 GT and the 6800 vanilla. All three of those are effected, its that simple. Nv45 is possibly different than Nv40 (or is it Nv40 with an outside of the core bridge?) but if its just a bridge to PCIe and not in the core then the flaw is still there.
Deathlike2
10-Dec-2005, 20:45
6800 vanilla is not affected... for whatever the reason...
The NV45 is just an NV40 that uses a different slot.. that's really all it is. (I don't know/care whether it uses the AGP->PCI-E bridge, but I haven't seen anything about this that has affected benchmarks).
Skrying
10-Dec-2005, 20:58
Uhh, BS. Yes it is too, or else my 6800 (two now, I just got an RMA back) is to affected. What would make it special? Its Nv40, what would make the vanilla 6800s special?
Also, the fact that either Nv45 uses a bridge or not is important to this arguement, not about any type of performance benchmarks. If its a bridge then I highly doubt they changed anything in the core to fix this flaw. If its not a bridge then its highly likely they did fix this probably in the core.
Deathlike2
10-Dec-2005, 21:10
Uhh, BS. Yes it is too, or else my 6800 (two now, I just got an RMA back) is to affected. What would make it special? Its Nv40.
Well, assuming you are using a 6800 vanilla, the reason I recall why it is not affected was that it was made AFTER the 6800GT/Ultra (the high end stuff comes out first of course)..
It doesn't really matter though... there's always some huge flaw that never seems to get enough attention.
FX - poor PS2 performance
6800 - borked WMV 9 decode acceleration
7800 - shimmering sensitive
Murakami
10-Dec-2005, 22:45
I didn't see the link... apparently NVidia made a mistake in their listing..
NV40 (Geforce 6800 Ultra/GT AGP) and NV45 (Geforce 6800 Ultra/GT PCI-E) are the hardware with the bug... I'll cut them a "little slack" as their tables are huge.. though it should be their job to get the info correct.
Did I not say that new drivers won't fix the FLAW in the hardware? You will not get any performance benefits if the hardware is not able to support it...
This is the link (http://www.nvidia.com/page/purevideo_support.html), already posted in this very thread; you already said that new drivers won't fix the flaw in the hardware, but you have to prove that a flaw exists, because nVidia says that NV45 video processor is fully functional and the only review that seems to deny this claim is very old.
Another link for you (http://www.beyond3d.com/previews/nvidia/nv45/): Although this chip is labelled as NV45, other than the on-package integrated bridge chip, it’s not clear what changes are in the chip. NVIDIA suggest there are some improvements... :twisted:
Well, assuming you are using a 6800 vanilla, the reason I recall why it is not affected was that it was made AFTER the 6800GT/Ultra (the high end stuff comes out first of course)..
It doesn't really matter though... there's always some huge flaw that never seems to get enough attention.
FX - poor PS2 performance
6800 - borked WMV 9 decode acceleration
7800 - shimmering sensitive
6800 vanilla can use different cores. NV40/45 are broken, NV41/42 not.
Murakami
10-Dec-2005, 23:02
6800 vanilla can use different cores. NV40/45 are broken, NV41/42 not.
Prove that NV45 are broken, please, and i'll thank you.
Skrying
10-Dec-2005, 23:31
6800 vanilla can use different cores. NV40/45 are broken, NV41/42 not.
I thought Nv41 and Nv42 were only used on the PCIe versions? Which from my understanding are not borked. Nv40 was used for the AGP versions correct?
Prove that NV45 are broken, please, and i'll thank you.
http://techreport.com/onearticle.x/7535
NV45 is NV40 + bridge.
Well, assuming you are using a 6800 vanilla, the reason I recall why it is not affected was that it was made AFTER the 6800GT/Ultra (the high end stuff comes out first of course)..
Initial shipments of 6800 vanilla were NV40 (binned cores) that didnt have 4 working quads, hence there was 50-50 chance of unlocking them.
Later on Nvidia used NV41 for the same.
Skrying
11-Dec-2005, 00:55
Initial shipments of 6800 vanilla were NV40 (binned cores) that didnt have 4 working quads, hence there was 50-50 chance of unlocking them.
Later on Nvidia used NV41 for the same.
Hmm, weird. I just go a RMA back from BFG (5 day RMA at that... amazing service), its not my old one fixed or anything, totally different heatsink and slightly different positioning of molex connector. But this one still works just fine, maybe BFG only ordered Nv40s so they could easily do the OC?
Deathlike2
11-Dec-2005, 06:10
Prove that NV45 are broken, please, and i'll thank you.
It doesn't have to be proven when every major website you can think of has repeated this major bit of info.. I'm not wasting my time repeating myself when a quick google search can answer your question..
One other thing..
Although this chip is labelled as NV45, other than the on-package integrated bridge chip, it’s not clear what changes are in the chip.NVIDIA suggest there are some improvements,however it may be the case that the chip itself is more akin to a respun NV40.Testing should indicate if there are any significant performance differences from the rest of the 6800 line.
Italics highlight the quoted, bold highlight the MOST IMPORTANT part of Dave's statement.
In any case, there's nothing particularly special about the NV45 (PCI-E version of the NV40). If you had paid any attention to all the websites providing the news, you would have already know this to be fact.. especially when some have quoted NVidia the information necessary to figure out what was going on.
I don't understand why this "old news" is that special.. given that this has been repeated everywhere... the only issue is probably NVidia's site documenting the info incorrectly (it was correct prior to the G70).
Murakami
11-Dec-2005, 08:11
Every major website i can think of were speculating, not proving anything with real tests.
In your opinion, the most important part of Dave's statement is "it may be the case"...my english is basic but "it may be the case" means only a possibility, whereas nVidia suggests "there are some improvements" and his support page lists NV45 as supporting WMV9 decode acceleration...maybe the improvements are exactly in video processor.
I'm sorry, but i'll stay with nVidia until someone can prove that his statement is plain wrong; this "old news" is so special for me because i bought a 6800 GT PCI-E exactly for his video processor... :???:
I'm going to send a mail to nVidia with this question... :twisted:
EDIT: i'm not able to find a valid e-mail address... :cry:
Deathlike2
11-Dec-2005, 08:28
Ugh..if you even READ the Anandtech link I posted.. it does exactly what you were asking for... though the problem is that they did not specify (it's late, so I didn't thoroughly check) which video card (PCI-E or AGP version) they used for testing.
Do yourself a favor.. scour the NVidia and NVNews forums and see what people are saying about Purevideo support in the PCI-Express versions of the 6800 (NV45)... then you will understand the truth of the matter. Your blind ignorance in not looking through forums (let alone the "speculation" as you call it, especially when reported by reputable sources) to at least see what other people have to say in the matter... that's all I will ask you to do..
Murakami
11-Dec-2005, 08:46
Ugh..if you even READ the Anandtech link I posted.. it does exactly what you were asking for... though the problem is that they did not specify (it's late, so I didn't thoroughly check) which video card (PCI-E or AGP version) they used for testing.
Do yourself a favor.. scour the NVidia and NVNews forums and see what people are saying about Purevideo support in the PCI-Express versions of the 6800 (NV45)... then you will understand the truth of the matter. Your blind ignorance in not looking through forums (let alone the "speculation" as you call it, especially when reported by reputable sources) to at least see what other people have to say in the matter... that's all I will ask you to do..
I'm tired to argue with you, our points of view are opposite: i'm not interested in forums' talk (that i've already read), speculation or something like that...i'm searching for a proof, an official statement, because my brain tells me that if nVidia admits that NV40 video processor is "borked" there were no problems admitting the same thing for NV45.
The difference between us is that you're losing your patience insulting me about my "blind ignorance", me not.
I'll do myself a favor: i'll test this infamous WMV9 acceleration feature on my own...the only real problem is that one of the 2 WMP10 patches that enable DXVA acceleration is not removable and i don't want to format my O.S. for that matter.
Deathlike2
11-Dec-2005, 08:56
I'm tired to argue with you, our points of view are opposite: i'm not interested in forums' talk (that i've already read), speculation or something like that...i'm searching for a proof, an official statement, because my brain tells me that if nVidia admits that NV40 video processor is "borked" there were no problems admitting the same thing for NV45.
The difference between us is that you're losing your patience insulting me about my "blind ignorance", me not.
I'll do myself a favor: i'll test this infamous WMV9 acceleration feature on my own...the only real problem is that one of the 2 WMP10 patches that enable DXVA acceleration is not removable and i don't want to format my O.S. for that matter.
Um.. even though the FX was a POS hardware for its generation.. NVidia admitted this years later when the 6800 series fixed the bad PS2 performance problem inherent in the FX architecture.
For the 6800 series.. it is still a staple selling point for the mid-high range.. and getting ANY sort of official response from NVidia will not happen anytime soon. No company will admit to bad products unless that product is well on its way out. This is no different. Official statements HAVE come out... but not from NVidia to the public directly, but only to the reviewers such as Anandtech who are relaying this info. You make it sound like it is speculation.. when really, it has come from the horse's mouth.. under someone else's voice (the reviewer in this case).
I'll do myself a favor: i'll test this infamous WMV9 acceleration feature on my own...the only real problem is that one of the 2 WMP10 patches that enable DXVA acceleration is not removable and i don't want to format my O.S. for that matter.
Lots of stuff done on your system is not exactly "easily" removable such as DX9 (though there is a tool to solve that, but this concept has always applied for many patches/software MS has provided)... I don't understand what is stopping you from testing it yourself given that you seemingly are very much interested in the results of the test. You really do need to test for yourself to know the truth.
Murakami
11-Dec-2005, 09:39
For a wise testing, i need to format my PC, install everything minus 2 WMP10 patches (DRM and DXVA), install Forceware 81.95 and monitoring CPU % (and, probably, GPU temp) during WMV9 playback, then install the patches and repeat...IMHO, this is the only way to be sure that NV45 video processor does not accelerate WMV9 decoding, because, unlike Ati drivers, nVidia's one haven't got a tool to switch on and off WMV9 acceleration...oh, and do the same things with a 6600 GT... :lol:
Maybe, one day... :roll:
Deathlike2
11-Dec-2005, 09:53
For a wise testing, i need to format my PC, install everything minus 2 WMP10 patches (DRM and DXVA), install Forceware 81.95 and monitoring CPU % (and, probably, GPU temp) during WMV9 playback, then install the patches and repeat...IMHO, this is the only way to be sure that NV45 video processor does not accelerate WMV9 decoding, because, unlike Ati drivers, nVidia's one haven't got a tool to switch on and off WMV9 acceleration...oh, and do the same things with a 6600 GT...
Maybe, one day...
Actually.. you don't have to go through the many steps you have suggested.
There are three parts to enabling WMV acceleration for NVidia cards...
Getting the Purevideo enabled drivers...
WMP patches of course...
The third part is enabling the feature in WMV9. If I recall correctly, the WMV9 acceleration MUST be enabled in WMP9 as well in order for it to work.
So, in order to test it.. all you have to actually do is turn the acceleration off in WMP9 and test... and then do a test with the acceleration on. WMV acceleration is three sided... WMP patch... driver support... then the app. Since you have control of the acceleration via WMP9.. you can test it this way.
Just a side note: The tipoff as the why the NVidia Purevideo support page is WRONG (and not totally correct).. is because the NV41 (PCI-E Geforce 6800 vanilla) is reported to not have the problem.
Murakami
11-Dec-2005, 10:01
I have WMP 10 (not removable without a system restore; and i turned off this feature before installing it... :lol: ) and i'm not able to find any setting about WMV9 acceleration: can you enlighten me? :shock:
P.S. The 2 patches are for WMP 10, not 9... :???:
Deathlike2
11-Dec-2005, 10:05
Hints:
http://www.shacknews.com/ja.zz?comments=37332
Look at post #4 from Carnivac..
Start at Step 7 assuming you have installed drivers that support Purevideo and installed whatever MS requires you.
9. Ensure Use high quality mode is selected. Click OK, OK.
This is the option to enabled the acceleration for whatever it is supposed to accelerate...
Murakami
11-Dec-2005, 10:19
Hints:
http://www.shacknews.com/ja.zz?comments=37332
Look at post #4 from Carnivac..
Start at Step 7 assuming you have installed drivers that support Purevideo and installed whatever MS requires you.
This is the option to enabled the acceleration for whatever it is supposed to accelerate...
Ok, i'll try; in my opinion, this option switches between hardware overlay support and VMR...maybe, VMR is required to engage (i love this word ala Star Trek... :oops: ) WMV9 acceleration... :cool:
Murakami
11-Dec-2005, 10:32
Tests done! :grin:
My system is a Venice 3000+; sample movie is T2_720.vmw.
Well, there's something strange: without acceleration CPU % stays between 30% and 55%, with acceleration between 30% and 45%...but the GPU temp is always 55 C°, no matter if "high quality mode" is set to on or off... :???:
For a comparison, during MPEG2 decoding (with PureVideo) GPU temp goes up to 60 C°... :???:
Deathlike2
11-Dec-2005, 14:05
Tests done! :grin:
My system is a Venice 3000+; sample movie is T2_720.vmw.
Well, there's something strange: without acceleration CPU % stays between 30% and 55%, with acceleration between 30% and 45%...but the GPU temp is always 55 C°, no matter if "high quality mode" is set to on or off... :???:
For a comparison, during MPEG2 decoding (with PureVideo) GPU temp goes up to 60 C°... :???:
So.. what do the results tell you?
On the simple basis, if the GPU temp is not going up.. between the mode on or off.. there is some acceleration going on.. but it is of the CPU kind (SSE/SSE2/SSE3)... you would think the GPU do work yes? Obviously this is not the case. Since you did a test in which know the GPU should do some work (as a result, seeing higher GPU temp)... you can at least see something going on.
Do you think your little experiment is "conclusive" enough? I believe so.
Murakami
11-Dec-2005, 20:07
Probably you're right, but...incidentally, a friend of mine (with the same software setup) was not able to measure any temp range during WMV playback.
Incidentally, this guy has a 6600 GT.
Murakami, he may be right but the temperature thing doesn't mean much at all. If the wm acceleration is only reliant on a very small section of the die then it is unlikely to really increase the temp at all. A more meaningful way would be to look at your friends and see what % increase he saw, if you see a similar increase with similar CPU then it is likely yours is working.
Murakami
11-Dec-2005, 22:46
Murakami, he may be right but the temperature thing doesn't mean much at all. If the wm acceleration is only reliant on a very small section of the die then it is unlikely to really increase the temp at all. A more meaningful way would be to look at your friends and see what % increase he saw, if you see a similar increase with similar CPU then it is likely yours is working.
Of course i'll try, but it would be very strange if WMV acceleration was working and GPU temp did not grow up even 1 C°, when MPEG 2 decoding goes up the temp 5 C° at least... :???:
However, it's hard to monitor CPU % only through task manager, it's too much subjective (given a small difference between software decoding and hardware assisted one): any suggestion?
Murakami
11-Dec-2005, 22:58
Uh uh, i found this (http://www.beyond3d.com/previews/nvidia/g70/index.php?p=21): WMV9 decoding, G70 next to 6800 Ultra PCI-E...almost a draw! :shock:
Deathlike2
12-Dec-2005, 03:08
Probably you're right, but...incidentally, a friend of mine (with the same software setup) was not able to measure any temp range during WMV playback.
Incidentally, this guy has a 6600 GT.
I would suggest asking him to run the simple test that I was discussing about earlier.. and see the CPU utilization... though the 6600 series (all varients) have no WMV9 decode acceleration issues.
Uh uh, i found this: WMV9 decoding, G70 next to 6800 Ultra PCI-E...almost a draw!
Well, according to Dave's opinion on the results.. he believes there are one of two things possibly at work:
1) little or no improvement with the Purevideo decoding
2) unoptimized drivers
I believe it is probably closer to the latter (since there hasn't been any reports of the 7800 series having the same hardware flaw)... It would be a very stupid thing indeed if the same flaw has not been addressed in the G70...
Murakami
12-Dec-2005, 07:43
I would suggest asking him to run the simple test that I was discussing about earlier.. and see the CPU utilization... though the 6600 series (all varients) have no WMV9 decode acceleration issues.
Well, according to Dave's opinion on the results.. he believes there are one of two things possibly at work:
1) little or no improvement with the Purevideo decoding
2) unoptimized drivers
I believe it is probably closer to the latter (since there hasn't been any reports of the 7800 series having the same hardware flaw)... It would be a very stupid thing indeed if the same flaw has not been addressed in the G70...
You're kidding, right? First, you suggest me to look at GPU temp variation, assuming that CPU% difference i misured is due to software optimisation in high quality mode (MMX, SSE and so on)...second, you suggest a friend of mine to look at CPU% difference and not GPU temp variation... :lol: Please, a little of coherence! :roll:
Besides, 7800 GTX preview i linked can illustrate one simple thing: 6800 Ultra PCI-E has no hardware flaw (hardware flaw that this review does not mention; i suppose that, if present, the reviewer chose a 6600 GT for the comparison).
Deathlike2
12-Dec-2005, 07:56
You're kidding, right? First, you suggest me to look at GPU temp variation, assuming that CPU% difference i misured is due to software optimisation in high quality mode (MMX, SSE and so on)...second, you suggest a friend of mine to look at CPU% difference and not GPU temp variation... Please, a little of coherence!
Did I say ANYTHING about testing the GPU temp variation? I only made an analysis based on what I understood from the results (which includes what you said about the GPU temps). I didn't say you test your GPU temps.. just the CPU utilization.
I hope you at the very least reread what I said.
Murakami
12-Dec-2005, 08:34
Did I say ANYTHING about testing the GPU temp variation? I only made an analysis based on what I understood from the results (which includes what you said about the GPU temps). I didn't say you test your GPU temps.. just the CPU utilization.
I hope you at the very least reread what I said.
"On the simple basis, if the GPU temp is not going up.. between the mode on or off.. there is some acceleration going on.. but it is of the CPU kind (SSE/SSE2/SSE3)... you would think the GPU do work yes? Obviously this is not the case. Since you did a test in which know the GPU should do some work (as a result, seeing higher GPU temp)... you can at least see something going on"
No comment; you're good to play with words but, if i have to look only at CPU utilization and not GPU temp, i have to conclude that, basing on my tests, my VGA does assist WMV9 decoding in hardware.
Probably you're right, but...incidentally, a friend of mine (with the same software setup) was not able to measure any temp range during WMV playback.
Incidentally, this guy has a 6600 GT.
Your friend might not be getting hardware accel.
Look at the clockspeed graph with rinatuner.. it should clock to to 3d speeds :D
Murakami
12-Dec-2005, 09:02
Your friend might not be getting hardware accel.
Look at the clockspeed graph with rinatuner.. it should clock to to 3d speeds :D
Excellent: so, no even a 6600 GT does WMV9 hardware decoding... :lol:
Besides, are you sure that the clock rises during 3D and/or video decoding?
Another test:
http://www.xbitlabs.com/articles/video/display/radeon-x1000_38.html
Murakami
12-Dec-2005, 10:40
Another test:
http://www.xbitlabs.com/articles/video/display/radeon-x1000_38.html
Good: the differences between 6800 Ultra and 7800 GTX are minimal.
Furthermore: "By the first GeForce 6800 Ultra modifications with disabled video processor, the CPU Utilization level could be as high as 100%"... :grin:
mrcorbo
15-Dec-2005, 04:32
Good: the differences between 6800 Ultra and 7800 GTX are minimal.
Furthermore: "By the first GeForce 6800 Ultra modifications with disabled video processor, the CPU Utilization level could be as high as 100%"... :grin:
Sorry about being away from this thread, kinda thought it had died....
Keep in mind that they were using a 3.6GHz PIV in their test. That's going to be able to handle the WMV-HD decoding just fine with or without assist from the GPU. It's possible that the 6800 Ultra's results were, in fact, showing the result of the broken VPP (it was slower, even if the margin wasn't that large).
Anyway you wanted a link to something official? Here it is, right from the horses mouth Nvidia customer support:
http://nvidia.custhelp.com/cgi-bin/nvidia.cfg/php/enduser/popup_adp.php?p_faqid=108&p_created=1098889190
In all honesty, if you've got a 6800-class vid card in your machiine you are also very likely to have it paired with a CPU that's capable of playing back these clips without any hardware assist, so it's not really that big of a deal. Still, NVidia should be accurate on their own website. For the record, I think it is an error and not a deliberate attempt to mislead. If you look at the tables you might notice another error, too. It shows the page designer didn't really proofread the page very well.
Murakami
15-Dec-2005, 10:00
:sad: :mad: :cry:
Do somebody have a link to an MPEG2-HD clip?
mrcorbo
15-Dec-2005, 18:52
:sad: :mad: :cry:
Do somebody have a link to an MPEG2-HD clip?
Try this page:
http://www.fusionhdtv.co.kr/ENG/Download/Demo.aspx
There are some sample clips available there.
Deathlike2
15-Dec-2005, 21:17
So.. was this "information" a bunch of BS?
If this information was false, wouldn't NVidia make an attempt to disclaim this info? The fact the Anand and other websites HAD IT RIGHT... choosing not to follow what NVidia had told them is just... :roll:
Though, I'm disappointed at the minor fact that there hasn't been a lot of benching of CPU utilization for Purevideo (and soon to come AVIVO) HD video encoding/decoding...
Murakami
14-Jan-2006, 15:45
Pure Video supporto page has changed: now, 6800 GT/Ultra PCI-E does not support WMV9 acceleration...:evil:
I immediately changed my 6800 GT with a 6600 GT...:twisted:
Murakami
14-Jan-2006, 22:36
New tests with "T2: 720p" and my brand new 6600 GT...nothing changed: CPU % is the same (with or without "high quality mode" switch in WMP10, 2 patches already installed), GPU temp does not rise...it doesn't work! :shock:
Deathlike2
14-Jan-2006, 22:56
Did you remember to install the Purevideo decoder? That is required (it's not just a driver thing)..
http://www.nvidia.com/object/nzone_purevideo_requirements.html
Murakami
14-Jan-2006, 23:01
Did you remember to install the Purevideo decoder? That is required (it's not just a driver thing)..
http://www.nvidia.com/object/nzone_purevideo_requirements.html
Yes, i installed PureVideo decoder, but PureVideo is only an MPEG decoder: in fact, it shows its icon in the systray only if WMP plays a DVD.
Deathlike2
15-Jan-2006, 00:08
Considered reinstalling the drivers+Purevideo decoder after changing the video card? It might be worth trying here.
Yes, i installed PureVideo decoder, but PureVideo is only an MPEG decoder: in fact, it shows its icon in the systray only if WMP plays a DVD.
erm purevideo does wmv-hd steams along other things.
If you have purevideo installed you should also get the purevideo systray item.
Murakami
15-Jan-2006, 07:21
erm purevideo does wmv-hd steams along other things.
If you have purevideo installed you should also get the purevideo systray item.
No, PureVideo Decoder is only an MPEG decoder: http://www.nvidia.com/object/dvd_decoder_techspecs.html
http://www.nvidia.com/object/decoder_faq.html
"The NVIDIA DVD Decoder combines the industry’s highest quality DVD and MPEG-2 playback with rich surround sound and is capable of supporting any Microsoft DirectX 9 compatible graphics product. Once installed, no additional setup is required. The DVD Decoder runs automatically whenever MPEG-2 or DVD video is launched in Windows Media Player. " from http://www.nzone.com/object/nzone_purevideo_requirements.html
Murakami
15-Jan-2006, 07:22
Considered reinstalling the drivers+Purevideo decoder after changing the video card? It might be worth trying here.
I've just formatted my system with no luck.
No, PureVideo Decoder is only an MPEG decoder: http://www.nvidia.com/object/dvd_decoder_techspecs.html
http://www.nvidia.com/object/decoder_faq.html
Well.. wmv-hd is a feature of purevideo according to nvidia, what you're referring to is the dvd decoder that you pay for.
http://nvidia.com/page/purevideo.html
pureVideo delivers:
High-Definition Video: NVIDIA PureVideo technology finally allows you to experience lifelike HD video on your PC, notebook, or HDTV without the expense of additional home-theater devices. With dedicated hardware to accelerate MPEG-2/DVD as well as the new Microsoft® Windows Media® HD Video (WMV HD) standard, your DVDs come to life in detail never before possible on a PC. Using NVIDIA PureVideo technology, the GPU offloads video decoding from the CPU, resulting in smooth, stutter-free, high-definition video playback. And, since PureVideo supports most current and future HD formats, you can rest assured that a PC powered by PureVideo will keep up with the pace of tomorrow’s video technologies, and continue to deliver HD content well into the future.
Murakami
15-Jan-2006, 07:29
Well.. wmv-hd is a feature of purevideo according to nvidia, what you're referring to is the dvd decoder that you pay for.
http://nvidia.com/page/purevideo.html
Exactly what i mean: PureVideo is the technology behind the video processor, PureVideo decoder is the MPEG decoder. To get WMV-HD acceleration support, PureVideo decoder is not required; only Forceware drivers, WMP10 and two WMP patches, just like Ati.
Exactly what i mean: PureVideo is the technology behind the video processor, PureVideo decoder is the MPEG decoder. To get WMV-HD acceleration support, PureVideo decoder is not required; only Forceware drivers, WMP10 and two WMP patches, just like Ati.
Oh I see what you meant.
ya.. you dont need to pay to get wmv-hd accel.
Murakami
15-Jan-2006, 08:09
Oh I see what you meant.
ya.. you dont need to pay to get wmv-hd accel.
Yes: i don't need to pay because it does not work... :lol:
Yes: i don't need to pay because it does not work... :lol:
works for 6600GTs unless your card is borked.
Murakami
15-Jan-2006, 08:34
Well, there's something strange here: unlike 6800GT, during MPEG2 playback (with PureVideo decoder) GPU temp rises only by 1° C...exactly like WMV-HD playback... :shock:
Switching to "overlay mode" in WMP10 does not change anything during WMV9 playback: in order to test CPU %, i have to uninstall the two WMP10 patches.
Murakami
15-Jan-2006, 09:06
Ok, i uninstalled the patches and CPU % goes up to 40-50 degree; reinstalled the patches and CPU % goes down to 20-30 degree.
Conclusion: NV43 WMV-HD acceleration does work; GPU temp is not a reliable indicator (6600 GT related only).
Ok, i uninstalled the patches and CPU % goes up to 40-50 degree; reinstalled the patches and CPU % goes down to 20-30 degree.
Conclusion: NV43 WMV-HD acceleration does work; GPU temp is not a reliable indicator (6600 GT related only).
As I said much earlier in thread, you should be looking at gpu speed in rivatuner.. it goes to 3d clocks when accerating video.
Murakami
15-Jan-2006, 09:51
As I said much earlier in thread, you should be looking at gpu speed in rivatuner.. it goes to 3d clocks when accerating video.
According to RivaTuner, GPU stays at 300 mhz during video playback.
According to RivaTuner, GPU stays at 300 mhz during video playback.
odd.
My BFG 6600GT would clock up to 525mhz (factory overclocked) during wmv-hds.
Murakami
15-Jan-2006, 10:00
odd.
My BFG 6600GT would clock up to 525mhz (factory overclocked) during wmv-hds.
Odd: mine not, no matter if WMV-HD or MPEG2 playback... :???:
I have an Asus Extreme N6600 GT; drivers are ForceWare 82.12... :mad:
Odd: mine not, no matter if WMV-HD or MPEG2 playback... :???:
I have an Asus Extreme N6600 GT; drivers are ForceWare 82.12... :mad:
Let me see if my 7800GT does:smile:
Nope.
So I wonder why my 6600GT did?
Murakami
15-Jan-2006, 10:23
Let me see if my 7800GT does:smile:
Nope.
So I wonder why my 6600GT did?
I don't know... :sad:
I don't know... :sad:
It bothers me that my last card ramped up and yours doesn't and my new card doesn't :evil:
Murakami
15-Jan-2006, 10:30
It bothers me that my last card ramped up and yours doesn't and my new card doesn't :evil:
Probably, some differences in VGA BIOS. RivaTuner reports my NV43 as "A4" revision: which is yours?
Probably, some differences in VGA BIOS. RivaTuner reports my NV43 as "A4" revision: which is yours?
no idea...
Murakami
15-Jan-2006, 10:58
no idea...
See what RivaTuner says to you... :shock:
See what RivaTuner says to you... :shock:
erm... I have a new system now..
6600GT along with AXP and mobo are in my closet waiting to be sold.
Murakami
15-Jan-2006, 13:03
erm... I have a new system now..
6600GT along with AXP and mobo are in my closet waiting to be sold.
Understood... :grin:
Understood... :grin:
Ya know what's sad?
I'm dissapointed with the serious sam 2 performance:sad:
Also it's true what people say about the geforce 7 texture filtering.
Murakami
16-Jan-2006, 08:26
Ya know what's sad?
I'm dissapointed with the serious sam 2 performance:sad:
Also it's true what people say about the geforce 7 texture filtering.
Odd: SS2 was very good with my old 9800 pro and perfect with a 6800 GT... :???:
Odd: SS2 was very good with my old 9800 pro and perfect with a 6800 GT... :???:
It does great outdoors, it's certain indoor places.
It has to do with the lighting I've discovered.
Outdoors I can do 1280x1024 4x fsaa 16AF no sweat.
vBulletin® v3.8.6, Copyright ©2000-2013, Jelsoft Enterprises Ltd.