HardOCP R520 benchmarks

T

tranCendenZ

2[H]4U
Joined
Jun 6, 2004
Messages
3,844
Brent/Kyle, was looking forward to the annual NV/ATI joust on HardOCP.

I was wondering, for your review would it be possible to include benchmarks with the new Forceware v80 drivers that are multithreaded? It seems this will be Nvidia's next big driver release and it can give 10-40% gains depending on the application. 81.26 is the latest driver available to the public but it has some old bugs in it... That being said, is it possible you could include the latest Rel.80 drivers NV provides you to give insight to those who have or plan to buy dual core systems? Maybe in addition to a 70 series driver if you don't get a bugfree 80 series driver? :D
 
Errrrr... Bret has always made it a point to test only with drivers available off their respective website and not guru3d.

Nice begging though.
 
R1ckCa1n said:
Errrrr... Bret has always made it a point to test only with drivers available off their respective website and not guru3d.
Click to expand...

Well, INQ claims that Rel.80 will be available on the website near (maybe the day before) R520 launch, the question is whether HardOCP will run some tests with it in addition to the old drivers in order to more accurately reflect current performance on NV cards. I remember when the 6800GT first came out one or two HardOCP reviews were not all that useful in showing performance differences because newer drivers with 20% performance gains had come out a week before the review showed up. Maybe just run some apples to apples tests with the new Rel80 since it would probably take too long to do the whole HardOCP method with them in addition to the 70 series.

Nice begging though.
Click to expand...

Thanks :D
 
Wow Tranny.... You are really reaching with the R520 coming very soon. First you quote penstarsystems and now theinq.

I'm running the 81.26 from guru3d and find my desktop doesn't like to redraw everytime......
 
R1ckCa1n said:
Wow Tranny.... You are really reaching with the R520 coming very soon. First you quote penstarsystems and now theinq.
Click to expand...

You really think Nvidia isn't going to try to spoil the R520 launch by suprising the competition with more competitive products as every videocard manufacturer attempts to do if they can? These drivers would be the perfect trump card, especially for dual core systems. I plan to get dual core as my next cpu configuration but I'd also be curious to see Rel80 7800GTX versus R520 on hyperthreaded P4.

And TBH, they are pretty important as they give the biggest performance gains even for 6800 series users in addition to 7800 series users since the Rel65 drivers!

I'm running the 81.26 from guru3d and find my desktop doesn't like to redraw everytime......
Click to expand...

They definitely do have bugs but if you notice the date on them, they are also relatively old drivers so odds are Nvidia has something newer and better ready to go. 81.26 wouldn't be a good set to benchmark with, but I would think Nvidia would release or provide a more polished set of 80-series drivers within the next week or two.
 
Of course you would want them used. They seem to be faster. I guess the fact that they are very buggy doesnt mean anything to you though.

[H] also seems to be using a FX-55 in its tests, not a X2. Perhaps they can go buy one of those for you too?
 
i am using the 81.26. desktop ok, gta ok, doom3 ok, bf2 ok, farcry ok. picked up nearly 10% in 3dmark. i dont know about 30-40% from drivers tho. seems a bit far fetched to me.

very truly yours,
politenessman


number 1, all cpus 3000 and under, all gpus, all drivers, all os's
http://service.futuremark.com/compare?3dm05=1246973
 
fallguy said:
Of course you would want them used. They seem to be faster. I guess the fact that they are very buggy doesnt mean anything to you though.

[H] also seems to be using a FX-55 in its tests, not a X2. Perhaps they can go buy one of those for you too?
Click to expand...

Sounds good to me :D
 
I'd only include them if they fix the weired texture issues in games until then. Otherwise it's more a marketing release than anything else IMHO. Just perfect to piss on ATIs parade :D
 
I can only say lol. :D
Tranc is scared that his beloved brand is outperformed by R520, and will grap every effort to
do some damage control. I'm sure that in case of Tranc was a reviewer, he would even use drivers which are only rendering black/white. :D Correct, buggyfree rendering is not important, only numbers! This is what the people in general look at.
 
perp said:
i am using the 81.26. desktop ok, gta ok, doom3 ok, bf2 ok, farcry ok. picked up nearly 10% in 3dmark. i dont know about 30-40% from drivers tho. seems a bit far fetched to me.

very truly yours,
politenessman


number 1, all cpus 3000 and under, all gpus, all drivers, all os's
http://service.futuremark.com/compare?3dm05=1246973
Click to expand...

Try running it at a res that someone would play a game at. Such as something else than 1024x768.
 
dont all drivers have bugs? isnt this the reason there are driver revisions? driver releases come with a list of "bug fixes". i always thought this pertained to the fact that the previous release had some bugs, this release fixes them, but i guess i am mistaken. apparently the 81.26 drivers have "extra" bugs, or somehow are buggy -er than other drivers. i demand a "bug pecentage". drivers can only have a certain percantage of bugs before they become officially buggy, and any drivers above the bug line can not be used in any comparison or even referred to in forums.

very truly yours,
politenessman
 
fallguy said:
Try running it at a res that someone would play a game at. Such as something else than 1024x768.
Click to expand...

righty right. i guess somehow you would know exactly what res i run my desktop and games at. also, i am sure you are running a nvidia product so would know exactly how these drivers would respond at hi res.

very truly yours,
politenessman
 
perp said:
dont all drivers have bugs? isnt this the reason there are driver revisions?
Click to expand...

Read the Nvnews forum about this driver, lots of people report rendering errors such as missing textures and not rendering shadows correctly or even no shadows at all.
 
oh boy.
my desktop runs fine. my games run fine. i got a 10% bump in 3dmark05. thats all i claim. if joe splivingates over at festerbestertester.com doesnt get shadows, i sympathize with him, and wish they ran as good for him as they do for me, and hope he has better results in the future. why do i somehow sense this is becoming another ati vs nvidia debate?


very truly yours,
politenessman

number 1, all cpus 3000 and below, all gpus, all drivers, all os's
http://service.futuremark.com/compare?3dm05=1246973
 
perp said:
righty right. i guess somehow you would know exactly what res i run my desktop and games at. also, i am sure you are running a nvidia product so would know exactly how these drivers would respond at hi res.

very truly yours,
politenessman
Click to expand...

Your desktop res has nothing to do with game res.

Are you really trying to tell me that with a A64 X2@2.8gig and 2x7800GT's that you run games at 1024x768?

Yes I have a NV card, a 7800GTX. Not that it matters.

The whole point was, that you are trying to judge these drivers on a synthetic benchmark, at 1024x768. Try a real game, at a real res. If the 10% is still there, GREAT! I just dont think it will be.

Yes I know 1024x768 is a CPU bench res today, however I really doubt many people with such systems play at that res.
 
fallguy said:
Are you really trying to tell me that with a A64 X2@2.8gig and 2x7800GT's that you run games at 1024x768?
Click to expand...
Where did he say anything about running his games at 1024x768?? Sheesh, the guy says the 81.26 driver runs fine for him and you're all over him with this 1024x768 fantasy of yours.
 
coz said:
Where did he say anything about running his games at 1024x768?? Sheesh, the guy says the 81.26 driver runs fine for him and you're all over him with this 1024x768 fantasy of yours.
Click to expand...
3d mark defaults to 1024x768....
 
tranCendenZ said:
would it be possible to include benchmarks with the new Forceware v80 drivers that are multithreaded?
Click to expand...
I'd love to see a finished Rel.80 driver tested on an X2 platform against R520. But I certainly don't want to see this crappy leaked Alpha version being used!!
 
coz said:
I'd love to see a finished Rel.80 driver tested on an X2 platform against R520. But I certainly don't want to see this crappy leaked Alpha version being used!!
Click to expand...
Given NV's track record with drivers over the past years, I would say not much will change from these drivers to the released ones.
 
fallguy said:
Your desktop res has nothing to do with game res.

Are you really trying to tell me that with a A64 X2@2.8gig and 2x7800GT's that you run games at 1024x768?

Yes I have a NV card, a 7800GTX. Not that it matters.

The whole point was, that you are trying to judge these drivers on a synthetic benchmark, at 1024x768. Try a real game, at a real res. If the 10% is still there, GREAT! I just dont think it will be.

Yes I know 1024x768 is a CPU bench res today, however I really doubt many people with such systems play at that res.
Click to expand...

i dont know what the hell you are talking about. i said that on MY machine the desktop was ok as were several games. i added that on MY machine i picked up 10% on 3dmark.
thats all. i made no blanket statements
you replied - well try that at this and this and that, because you have researched this thoroughly i guess.

now again you are making wacky ass assumptions on what i run my games at. no i am not trying to tell you anything except that "these drivers run ok on MY machine, the desktop is ok on MY machine and some games run ok on MY machine and i picked up 10% in 3dmark05" now if they dont do this well on yours, i sympathize, and hope you get better results in the future, or with other driver releases. thank you and goodnight.

very truly yours,
politenessman
 
Well I didn’t run it through a translator so I couldn’t read 90% of what was being said. However it is my opinion that those test results me nothing.

To begin with I have no idea what kind of standards they use for testing. All I can do is assume that all the same setting were used for each test, great. However when you’re already getting 60+ FPS average what good is another 20 FPS?

Sorry I will wait for the [H] to review this next generation of drivers to see what they are really worth. As it is my opinion there testing methods are superior. I am much more interested in gains that mean something like being able to bum the resolution up with the new driver while not having to drop AA settings and such to maintain playable frame rates. If those drivers are able to achieve such things with out sacrificing image quality then great! If not then those “massive” gains you mentioned mean nothing to someone like me who plays games rather then benchmarking them.


tranCendenZ said:
Massive gains on an X2 4800+:

http://www.hardwareoc.at/Geforce-7-Treibervergleich-5.htm
http://www.hardwareoc.at/Geforce-7-Treibervergleich-7.htm
Click to expand...
 
R1ckCa1n said:
Given NV's track record with drivers over the past years
Click to expand...

Correct me if I'm wrong, but wasn't nVidia involved in some "over-optimizing" issue as well? Can't erally remember though...
 
coz said:
I'd love to see a finished Rel.80 driver tested on an X2 platform against R520. But I certainly don't want to see this crappy leaked Alpha version being used!!
Click to expand...

Well, 81.26 are relatively old if you look at the driver date. I'd wager NV has a newer, less buggy Rel80 driver in the wings.
 
fallguy said:
Try running it at a res that someone would play a game at. Such as something else than 1024x768.
Click to expand...

Why ? Because 1024x768 is the standard resolution people use when benchmarking their comps.
 
wizzackr said:
Correct me if I'm wrong, but wasn't nVidia involved in some "over-optimizing" issue as well?
Click to expand...
Both ATi and NV's record when it comes to driver cheating or "over-optimizing" is far from spotless. I doubt anyone here can seriously suggest that either IHV is above such tactics.
 
Dammit, seeing the thread title with Brent Justice as the last poster in the forum summary page IS NOT FREAKING COOL IF THERE ARE NO BENCHMARKS AFTER ALL!
 
Morley said:
Dammit, seeing the thread title with Brent Justice as the last poster in the forum summary page IS NOT FREAKING COOL IF THERE ARE NO BENCHMARKS AFTER ALL!
Click to expand...

I don't "benchmark" anyway's, so.... ;)
 
coz said:
Both ATi and NV's record when it comes to driver cheating or "over-optimizing" is far from spotless. I doubt anyone here can seriously suggest that either IHV is above such tactics.
Click to expand...

Exactly - that's why I didn't understand the argument for why we should take these alpha drivers (which are about to be released right BEFORE a major ATI product launch) more serious than anything else, just because 'we know nVidia' certainly not.
 
Brent_Justice said:
I don't do benchmarks anyways, so.... ;)
Click to expand...
Yeah yeah, I know, I'm still adjusting my brain to use the term evaluation...after so many years of saying 'benchmarks'!
 
As for the original question asked about Rel 80.

We use officially available WHQL drivers UNLESS a company has officially made publicly available a beta driver that provides a performance or IQ difference. For example if NVIDIA posted a new driver on NZone that fixed a bug, or made performance or IQ better than the WHQL driver on NVIDIA.com. In that case I'll use the NZone driver because it is officially supported and publicly available through NVIDIA. I will not use LEAKED drivers that have ended up on Guru3D or whatever.

I use what is available at the time I start testing the cards.
 
Brent_Justice said:
As for the original question asked about Rel 80.

We use officially available WHQL drivers UNLESS a company has officially made publicly available a beta driver that provides a performance or IQ difference. For example if NVIDIA posted a new driver on NZone that fixed a bug, or made performance or IQ better than the WHQL driver on NVIDIA.com. In that case I'll use the NZone driver because it is officially supported and publicly available through NVIDIA. I will not use LEAKED drivers that have ended up on Guru3D or whatever.

I use what is available at the time I start testing the cards.
Click to expand...

That is fair, as recently the nZone drivers have tended to be superior to the official release drivers.
 
You must log in or register to reply here.
Back
Top