My benches so far. Adding GTX480 SLI benches.

[Janooo]

If you're serious and not just playing around, sure, we could try to guess why.
But all we really need to do is compare my rig to the one AT used. Find any differences, that could explain why are results aren't identical, just as other sites are not identical to AT's.

I had a look before I posted and the comps look similar. Not much difference.
I don't believe the rigs would explain moving the numbers in opposite directions. The same would go for settings. What else?

 

[Madcatatlas]

Do you keep files on everybody Keys?

 

[Lonyo]

If you're serious and not just playing around, sure, we could try to guess why.
But all we really need to do is compare my rig to the one AT used. Find any differences, that could explain why are results aren't identical, just as other sites are not identical to AT's.

It's not about being non-identical, it's about your results for the GTX480 being better than the AT ones while your AMD ones are lower than AT ones.

One might expect either both to be better or both to be worse, not one to be higher and one lower.

 

[happy medium]

Do you keep files on everybody Keys?

Na ,just the dip-shits. I do the same thing.

 

[nismotigerwvu]

Do you think if it were you, you would feel like doing what you suggest with the kind of comments you see from individuals in this thread? Sure, I have some resources, but my enthusiasm for sharing it has pretty much been extinguished. I mean, it took all my will just to do "this" many benches for AT. I could have done soooo much more, but what's the point really. It's fun for me, but not the constant bashing and being called a liar every 8.2 seconds. I don't think you'd like that very much either. So, I'm sorry.

Not to beat the dead horse much longer, but you seriously expected that you could post what you did and get nothing but positive feedback? Also, as far as the Stalker numbers look, it just seems like key's system is more cpu bound than the AT system. Perhaps the 400 series fares better in these situations. Might be interesting to see some numbers on lower clocked/lesser cored systems.

 

[happy medium]

What if you used a program to slow down the fan?
How much higher do the temps go?
Most GPU will run up to 120C before problems arise unless they are o/c or the vendor chose to raise clocks out the door for competitive edge and reduce the safety net, so to speak.

Ruby, I did read a review that stated that the gtx 480 fan on 60% kept the card at the same temp as on 100%, so there was no need to keep it that high. I believe they were overclocking also.

I read so many reviews there all jumbled up in my head.
If I can find it for you, I will post it.

 

[thilanliyan]

Ruby, I did read a review that stated that the gtx 480 fan on 60% kept the card at the same temp as on 100%, so there was no need to keep it that high. I believe they were overclocking also.

I seriously doubt that. Maybe that was at idle. If that were the case at load why have a fan that goes up that high? In the hardocp noise test, the fan went up to full blast when it got to about 95C and then the temp didn't go higher since the fan ramped up.

 

[nitromullet]

Not very impressive for the GTX 480 considering how much more gpu this card has on it compared to the 5870 and how long we waited for it. The saving grace here is the minimum fps which definitely favor NVIDIA. I'd like to see a FRAPS graph alongside the bar graphs to see how these min/max/avg fps are distributed before I put too much stock into that though.

Aside from that, I think the decision for most this generation is going to be based on things that aren't benchable like price, availability, and desired features. I don't even think TDP will play a huge role if the gap in price between the 5870 and GTX 480 decreases.

 

[cliftonite]

If you're serious and not just playing around, sure, we could try to guess why.
But all we really need to do is compare my rig to the one AT used. Find any differences, that could explain why are results aren't identical, just as other sites are not identical to AT's.

Intel i7860 @ 3.4GHz
ASUS Maximus III
8GB DDR3 1600
Sound Blaster X-Fi Supreme
400GB HDD
Windows 7 64-bit (all updates)

VS

CPU: Intel Core i7-920 @ 3.33GHz
Motherboard: Intel DX58SO (Intel X58)
Chipset Drivers: Intel 9.1.1.1015 (Intel)
Hard Disk: OCZ Summit (120GB)
Memory: Patriot Viper DDR3-1333 3 x 2GB (7-7-7-20)


What differences between those two could cause the difference? Don't see anything major...

 

[happy medium]

I seriously doubt that. Maybe that was at idle. If that were the case at load why have a fan that goes up that high? In the hardocp noise test, the fan went up to full blast when it got to about 95C and then the temp didn't go higher since the fan ramped up.

It's possible, those fans in that test were on auto.
I can set my fan to 50% and run furmark and it will go to about 68c ,or I can leave it on auto and it will ramp up past 50% and give me about the same temps.

Same with my Tuniq tower and my cpu I leave it on med speed when benching or high speed and the temps are just about the same mabe 1 or 2c.

I really doubt the 5870 needs 100% fan speed either.

Keys, you got the cards...give it a shot?

 

[thilanliyan]

Intel i7860 @ 3.4GHz
ASUS Maximus III
8GB DDR3 1600
Sound Blaster X-Fi Supreme
400GB HDD
Windows 7 64-bit (all updates)

VS

CPU: Intel Core i7-920 @ 3.33GHz
Motherboard: Intel DX58SO (Intel X58)
Chipset Drivers: Intel 9.1.1.1015 (Intel)
Hard Disk: OCZ Summit (120GB)
Memory: Patriot Viper DDR3-1333 3 x 2GB (7-7-7-20)


What differences between those two could cause the difference? Don't see anything major...

Only thing I can think of is maybe a bit more hard drive thrashing in Keys' comp because of the SSD used in AT's benches? Shouldn't be too bad if a couple of runs are done though.

 

[Lonyo]

Only thing I can think of is maybe a bit more hard drive thrashing in Keys' comp because of the SSD used in AT's benches? Shouldn't be too bad if a couple of runs are done though.

Would be interesting to see some comparisons of 1GB and 2GB ATI cards (when they eventually get benchmarked) on various HDD setups, to see how much of an impact 1GB more VRAM gives vs having a faster HDD for when that needs accessing.
That's the major difference between the two setups, but I've never seen anything look at something like that, and since the GTX480 has more RAM it would make sense it would have less chance of suffering due to it.

 

[Keysplayr]

Do you keep files on everybody Keys?

Only those very deserving. Would you like me to start one for you?

 

[Keysplayr]

Would be interesting to see some comparisons of 1GB and 2GB ATI cards (when they eventually get benchmarked) on various HDD setups, to see how much of an impact 1GB more VRAM gives vs having a faster HDD for when that needs accessing.
That's the major difference between the two setups, but I've never seen anything look at something like that, and since the GTX480 has more RAM it would make sense it would have less chance of suffering due to it.

AT did mention that the 1GB on the 5870 may be hindering the minimum framerates, but then adjusted their opinion when testing games that didn't saturate the frame buffer. So, I'm certain there are cases where a 2GB FB would help the 5870.

 

[Keysplayr]

Not to beat the dead horse much longer, but you seriously expected that you could post what you did and get nothing but positive feedback? Also, as far as the Stalker numbers look, it just seems like key's system is more cpu bound than the AT system. Perhaps the 400 series fares better in these situations. Might be interesting to see some numbers on lower clocked/lesser cored systems.

Actually, wouldn't my system be "less" CPU bound than AT's test system?
Unless memory bandwidth just affected Stalker (Dual channel vs. Triple channel).

The i7 860 Turbo's higher than the i7 920. So, I don't know which would be faster. We could check out some benches of when the i7 860 launched to compare with i7 920.

 

[Dark4ng3l]

Ok, sounds good. Now let all those who DO approve of this thread continue the conversation. You're disapproval is noted and I totally sympathize with your plight. Now forgive me, but I need to forward this text document I've been saving up, full of every quote addressed from you to me since we first knew each other existed, to the moderation.

I think they'll find something a bit peculiar.
Again, I'm sorry, but you leave me little choice.

Wow that has me shaking in my boots... But seriously the main point I wanted to make in this thread is that since you have the card and can use it now that its not even out yet you could have done some tests that were not included in most reviews (ie overclocking and/or whether or not the fan HAS to run at 100% full load) instead of just giving us more of the same.

The truth is everyone has benchmarks and some of them will differ significantly. And when yours show the opposite of what Anand posted you have to wonder what caused that. different version of the game maybe or different drivers. It still looks suspect but benchmarks are benchmarks they vary by system.

The only thing that I see is that you have some sort of agreement that you can't overclock the thing or share any results of that with us or you would have at least given us some general information like ''expect about 15% core and 10% memory O/C and my max temps was 98c with those settings'' but instead you give us nothing but benchmarks that contradict what was posted on Friday?

 

[happy medium]

Wow that has me shaking in my boots... But seriously the main point I wanted to make in this thread is that since you have the card and can use it now that its not even out yet you could have done some tests that were not included in most reviews (ie overclocking and/or whether or not the fan HAS to run at 100% full load) instead of just giving us more of the same.

The truth is everyone has benchmarks and some of them will differ significantly. And when yours show the opposite of what Anand posted you have to wonder what caused that. different version of the game maybe or different drivers. It still looks suspect but benchmarks are benchmarks they vary by system.

The only thing that I see is that you have some sort of agreement that you can't overclock the thing or share any results of that with us or you would have at least given us some general information like ''expect about 15% core and 10% memory O/C and my max temps was 98c with those settings'' but instead you give us nothing but benchmarks that contradict what was posted on Friday?

Why not just ask nicely and move on? Why try to trash someones hard work? I don't see any advertisements next to his benchies? He did this for the guys here in our forum.

Weather he's a focus member or not doesn't matter to me or many others.
Many of us are grown up enough not to focus on the dumb shit and thread bashing and decide for ourselves what to read or what not to read.

There are plenty of fan-boy sites out there that give bias benches.
It's up to you what you read or believe.

Give it a rest dude. You started this shit at post #21 with your stalker comment. I would have gave you a vacation long ago if I were him.

 

[waffleironhead]

Wow that has me shaking in my boots... But seriously the main point I wanted to make in this thread is that since you have the card and can use it now that its not even out yet you could have done some tests that were not included in most reviews (ie overclocking and/or whether or not the fan HAS to run at 100% full load) instead of just giving us more of the same.

The truth is everyone has benchmarks and some of them will differ significantly. And when yours show the opposite of what Anand posted you have to wonder what caused that. different version of the game maybe or different drivers. It still looks suspect but benchmarks are benchmarks they vary by system.

The only thing that I see is that you have some sort of agreement that you can't overclock the thing or share any results of that with us or you would have at least given us some general information like ''expect about 15% core and 10% memory O/C and my max temps was 98c with those settings'' but instead you give us nothing but benchmarks that contradict what was posted on Friday?

You might consider toning down a bit. The powers that be have sanctioned the focus group members to be here and do what they do. So I guess you either need to deal with it or move on. Derek posted a disclaimer about the focus group awhile back that you may want to read.
In this thread here http://forums.anandtech.com/showthread.php?t=191819
specifically this post http://forums.anandtech.com/showpost.php?p=25978475&postcount=37

NVIDIA Focus Group Members cannot have this well rounded perspective unless they work very very hard to overcome the natural barriers to neutrality put in front of them.

And it's not even really their "fault" -- it is their perspective that is tainted.

I think Keys is doing a good job of trying to overcome the bias his position presents. Should you be skeptical? Sure. Do you need to constantly hound him? No.

 

[Lonyo]

Actually, wouldn't my system be "less" CPU bound than AT's test system?
Unless memory bandwidth just affected Stalker (Dual channel vs. Triple channel).

The i7 860 Turbo's higher than the i7 920. So, I don't know which would be faster. We could check out some benches of when the i7 860 launched to compare with i7 920.

The speed bin is a couple of clocks higher, although it might depend on how many cores STALKER uses. Then there's the 70MHz advantage anyway (and maybe different memory timings?)
If it is higher clocks giving better results on the NV system with the slower HDD resulting in worse performance from the ATI due to the smaller framebuffer it would make sense.

Anyone know of any STALKER CPU benchmarks?

 

[Madcatatlas]

In Dark4angels (why couldnt you just pick an easier name huh!?) defence, he is asking critical questions and pointing out why.
As others have in this topic.

Now, anyone care to answer moi questionè?
Why is 5870 beating 480 in 3dmark Vantage?

 

[thilanliyan]

Why is 5870 beating 480 in 3dmark Vantage?

I don't know the technical reason but I think even the 2900XT beat or was near the 8800GTX in 3DMark but lost in games so 3DMark is not a very good judge of how a card will perform in games.

 

[Keysplayr]

I don't know the technical reason but I think even the 2900XT beat or was near the 8800GTX in 3DMark but lost in games so 3DMark is not a very good judge of how a card will perform in games.

Seconded. ATI seemed to score better in Futuremark benches, but in the 2900XT's case, didn't really translate into real world performance. This is pretty much the very reason why I didn't bother with Vantage. Doesn't really tell any significant story.

 

[Janooo]

If you're serious and not just playing around, sure, we could try to guess why.
But all we really need to do is compare my rig to the one AT used. Find any differences, that could explain why are results aren't identical, just as other sites are not identical to AT's.

I had a look before I posted and the comps look similar. Not much difference.
I don't believe the rigs would explain moving the numbers in opposite directions. The same would go for settings. What else?

Any luck in finding the reason for the discrepancy between your and AT tests?

 

[Lonyo]

Any luck in finding the reason for the discrepancy between your and AT tests?

Some possible reasons have already been discussed in this thread, but unfortunately Keys isn't in a position to go through all the permutations to test out some of the theories. That would require things he doesn't have like an SSD and an entire new mobo+CPU combination.

 

[Will Robinson]

Unbiased Keysplayr wrote:

What else do you need to know? I guess you are not interested in performance, well, that's really freaking cool as well. And it's a good thing, because that is what I was benching. Performance. Especially when things get cranked up.

Actually just reading AT,HardOCP and Bit Tech gave me that info.
I don't think I need to see the "revised" Focus group version.
Imagine that...