Fermi's lead over Cypress shrinks w/ new drivers

[railven]

Now you need to prove the canned benches correspond to actual game play.

What is the point if it is exact when it doesn't matches game play?

Also I would still like to know why the numbers reported by fraps don't correspond to the sample data.

I to am curious why his data isn't accurate.

The issue with real gameplay benches like these is the author's own agenda comes into play. Since the start it has been clear that Happy's stance on these benches are negative.

I wouldn't put it pass him to just stair at the floor for a few seconds just to spike one bench.

It would prove his claim.

Hey Happy, can you run some FarCry2 gameplay now with FRAPs, I'd like to see how that data correlates to your timed-demo. That article about timed demo vs gameplay is interesting.

 

[ViRGE]

Yep. I will also state this. The law of Thermodynamics does not change for HurleyBird just because he feels that it does.

A piece of equipment that uses a specific amount of power and produces a specific amount of heat will continue to do so regardless of the cooling solution used. The cooling setup is just a method of transfering heat from one place to another. Regardless of using water cooling or air cooling the amount of heat put into a room from a video card is CONSTANT. The rate is different but the amount is the same. All water cooling does over air cooling is transfer more heat away from the video card chip to the ambient environment faster than an air cooling solution. This allows for the video card to run consistently cooler but NOT the room. Meaning if a given room is unbearably hot with a couple of GTX480s in sli on an air cooling solution then that means a water cooled solution would have the same affect ON THE ROOM. The heat (energy) doesn't magically disappear because a person wants it to. It has to go somewhere. All water cooling does is transfer more heat faster. This means a room would get hotter sooner.

Just to be pedantic (and oh how I love doing so!) there is an exception to this.

One of the more interesting properties of silicon transistors is that leakage isn't a fixed quantity. It's variable with temperature, which is why chips are prone to thermal runaway to some extent. The hotter a chip, the more power it leaks, which in turn makes the chip hotter and leaks more power, etc.

So if you could keep the chip cool(er), you could keep leakage down, reducing the total power consumption of the card and ultimately the amount of heat it dissipates. A water cooling setup in theory would fit this scenario.

Now with that said, I don't believe anyone has done a proper analysis of GPUs, particularly not with TSMC's 40nm work. So I have no idea just how bad leakage is - it's said to be bad, but does going from 70C to 90C cause another 10W of leakage? 40W? Etc.

And of course there's the fact that the water cooling setup consumes power of its own, more so than just the fan on a stock air cooler. So it's entirely possible that the water cooler consumes more power on its own than what's saved by reduced leakage at lower temperatures.

In any case, fundamentally you're right: no matter what temperature you get the GPU all that heat is going in to the room anyhow. But it is possible to change the amount of heat dissipated with a different cooling solution.

 

[NoQuarter]

Whats that have to do with a test being accurate or inaccurate?

I just proved thats its inaccurate. End of story.
I have 27 more MODERN games on my drive and I'm sure the results would be the same.

When measuring performance ,I dont want something that's off by 6,8,10%.
I want something consistant and fair and most important as accurate as possible.

You did not prove they were inaccurate you only proved what the margin of error is. Whether they are inaccurate or not is a determination of whether the margin of error is too large to be taken seriously.

We often have to deal with margin of error, it doesn't invalidate the results. Even canned benchmarks have a margin of error of like 2%, if you keep rerunning the same canned bench you'll get slightly different results each time. That doesn't make them invalid and it doesn't make the results non-reproducable, it only means reproduced results will fall within a certain % of each other.

Gameplay results are much more informative to me. I accept that real world results will vary more each run. It's more important to me to find out this result though because I know nVidia and AMD both optimize heavily for common benchmarking games and especially for specific scenes and sequences for canned benches, and they don't reflect what you'll *actually* see while playing. I'm more interested in how the game runs than how long the driver team spent optimizing a specific scene.

 

[Lonyo]

What we really need is for BFG/Apoppin to play a good hour or so of a few games with different systems and then compare their real world results to those from benchmarks.

 

[busydude]

I see you misunderstood the point of my post. This isn't just about the 5870 and 480. The is about the cream of the crop of every generation release from ATI and Nvidia. MANY people who buy the absolute cream of the crop cards will also buy a second or third one for SLI/CF. That isn't my opinion. That is a fact. I would almost say that more people buying the cream of the crop will buy more than 1 rather than just buying 1. I'm not talking about mid-range or low-end cards here. I'm not even talking about the second in-line cards like the 470 and 5850.

If you are so sure of the fact, prove it. Show some stats and I'll stand corrected.

It is interesting you brought up micro-stuttering. Micro-stuttering is an issue that plagues multi-GPU setups.

Yeah, you just reiterated what i said.

Your thinking is back in 2006 or 2007. I say this because you talk about multi-GPU rigs being a niche market. Well, that really isn't the case anymore when dealing with enthusiasts and the cream of the crop cards. In case you haven't noticed, the review in this thread is comparing the cream of the crop cards. I'm not referencing mid-range or low-end cards. I also say this because micro-stuttering isn't nearly as bad as it used to be anymore with ATI's and Nvidia's latest offerings.

Sure its not as bad as ATI 3xxx days, but its an inherent issue with multi-GPU setups. This issue has not been fully solved as of now.

As far as my thinking goes, its just fine. I simply don't think the same way as you do.
I prefer better gaming experience than higher frame rates and multi-GPU's are too dependent on individual games to deliver results without faults and this is the reason there are very few reviews which deal with Multi-GPU's.

Lastly, I must sincerely ask you the question if you have or have had any 5xxx or 4xx series cards?

Why is this concerned with this discussion? I don't know if it makes you feel better/worse, but I have a 5xxx series card.

 

[Tempered81]

A piece of equipment that uses a specific amount of power and produces a specific amount of heat will continue to do so regardless of the cooling solution used. The cooling setup is just a method of transfering heat from one place to another. Regardless of using water cooling or air cooling the amount of heat put into a room from a video card is CONSTANT. The rate is different but the amount is the same. All water cooling does over air cooling is transfer more heat away from the video card chip to the ambient environment faster than an air cooling solution. This allows for the video card to run consistently cooler but NOT the room. Meaning if a given room is unbearably hot with a couple of GTX480s in sli on an air cooling solution then that means a water cooled solution would have the same affect ON THE ROOM. The heat (energy) doesn't magically disappear because a person wants it to. It has to go somewhere. All water cooling does is transfer more heat faster. This means a room would get hotter sooner.

I have to disagree with you on this point. If you have a video card running 80C for 10 hours on air cooling and another running 40C for 10 hours on water cooling, both running the same program, same load, in the same sized room, the ending ambients will not be the same, nor will they reach the same maximum. Saying the 40C watercooled card will heat the room to its maximum ambient sooner than the 80C aircooled card is incorrect. If it was correct, would I be correct saying a -140C card on liquid helium, dry ice, or liquid nitrogen running full load creates the same warmer ambients in a room as does an 80C card on air cooling and reaches that maximum sooner?? ... doesn't make any sense.

 

[jvroig]

I have to disagree with you on this point. If you have a video card running 80C for 10 hours on air cooling and another running 40C for 10 hours on water cooling, both running the same program, same load, in the same sized room, the ending ambients will not be the same, nor will they reach the same maximum. Saying the 40C watercooled card will heat the room to its maximum ambient sooner than the 80C aircooled card is incorrect. If it was correct, would I be correct saying a -140C card on liquid helium, dry ice, or liquid nitrogen running full load creates the same warmer ambients in a room as does an 80C card on air cooling and reaches that maximum sooner?? ... doesn't make any sense.

That's because those are not just PC cooling solutions, they are also "coolants" by themselves. Those things (liquid nitrogen, dry ice) affects the ambient temperature by themselves, they'll make your room cooler.

Fans, heatsinks, heatpumps/TEC, and water cooling solutions do not have that same cooling effect on your ambient temps. They just transfer heat from your CPU to outside of your case. The heat gets blown out no matter what you are using.

 

[tincart]

Simple solution to issues raised in this thread.

Canned time demo benchmarks give you a consistent report of the performance you can expect in... a canned time demo.

Real-world gameplay benchmarks will give you a less consistent report of the performance you can expect in... real-world gameplay.

Both of these methods will be "accurate" in the sense that they will reliably deliver information on the kind of performance you are benchmarking for.

If you are comfortable with the margin of error in the latter option, then it seems to be the superior solution in terms of being able to inform those who want to buy and use a video card to play games, not just a time demo.

 

[GaiaHunter]

I have to disagree with you on this point. If you have a video card running 80C for 10 hours on air cooling and another running 40C for 10 hours on water cooling, both running the same program, same load, in the same sized room, the ending ambients will not be the same, nor will they reach the same maximum. Saying the 40C watercooled card will heat the room to its maximum ambient sooner than the 80C aircooled card is incorrect. If it was correct, would I be correct saying a -140C card on liquid helium, dry ice, or liquid nitrogen running full load creates the same warmer ambients in a room as does an 80C card on air cooling and reaches that maximum sooner?? ... doesn't make any sense.

In that case there is change of state, which consumes energy by itself and isn't used to increase the temperature of the room.

 

[happy medium]

I to am curious why his data isn't accurate.

The issue with real gameplay benches like these is the author's own agenda comes into play. Since the start it has been clear that Happy's stance on these benches are negative.

I wouldn't put it pass him to just stair at the floor for a few seconds just to spike one bench.

It would prove his claim.

Hey Happy, can you run some FarCry2 gameplay now with FRAPs, I'd like to see how that data correlates to your timed-demo. That article about timed demo vs gameplay is interesting.

Here ya go.

Frames, Time (ms), Min, Max, Avg
3397, 60000, 42, 87, 56.617

FPS 55 58 62 49 52 50 51 55 59 66 67 65 63 66 64 63 58 65 64 69 77 87 58 51 51 52 52 53 48 43 50 60 70 71 55 43 42 43 43 47 48 49 45 46 46 45 47 50 52 59 61 61 51 51 60 66 63 65 70 65

Frames, Time (ms), Min, Max, Avg
3557, 60000, 13, 89, 59.283

FPS 57 58 63 59 58 60 64 63 60 62 60 70 66 70 62 57 53 53 54 50 46 51 54 56 54 51 53 53 47 49 56 59 45 46 40 45 45 46 48 55 58 85 76 83 83 84 85 84 85 87 88 89 88 88 84 14 25 25 25 24

Frames, Time (ms), Min, Max, Avg
3512 60000, 15, 241, 73.533

FPS 61 65 58 57 57 62 63 62 60 65 61 53 55 57 52 46 47 53 49 50 57 62 64 53 52 52 54 64 59 60 60 63 60 57 60 67 60 50 49 50 49 50 45 54 51 50 46 49 53 45 41 43 44 45 45 44 43 42 54 55

Did you say margin of error? This is ridiculous, I can't even make it come close to the same on all three runs. Most games don't even do the same thing over and over ,enemies are in different spots, different things blow up, more smoke. How can you guys not see this?

Doing benchmarks this way is way to unpredictable. You have to use the same atmosphere over and over to get a true benchmark.
Thats why they been doing it with a benchmark program for years.

No test are alike ,sometimes not even close.
I proved my point.

 

[NoQuarter]

Here ya go.

Frames, Time (ms), Min, Max, Avg
3397, 60000, 42, 87, 56.617

FPS 55 58 62 49 52 50 51 55 59 66 67 65 63 66 64 63 58 65 64 69 77 87 58 51 51 52 52 53 48 43 50 60 70 71 55 43 42 43 43 47 48 49 45 46 46 45 47 50 52 59 61 61 51 51 60 66 63 65 70 65

Frames, Time (ms), Min, Max, Avg
3557, 60000, 13, 89, 59.283

FPS 57 58 63 59 58 60 64 63 60 62 60 70 66 70 62 57 53 53 54 50 46 51 54 56 54 51 53 53 47 49 56 59 45 46 40 45 45 46 48 55 58 85 76 83 83 84 85 84 85 87 88 89 88 88 84 14 25 25 25 24

Frames, Time (ms), Min, Max, Avg
3512 60000, 15, 241, 73.533

FPS 61 65 58 57 57 62 63 62 60 65 61 53 55 57 52 46 47 53 49 50 57 62 64 53 52 52 54 64 59 60 60 63 60 57 60 67 60 50 49 50 49 50 45 54 51 50 46 49 53 45 41 43 44 45 45 44 43 42 54 55

Did you say margin of error? This is ridiculous, I can't even make it come close to the same on all three runs. Most games don't even do the same thing over and over ,enemies are in different spots, different things blow up, more smoke. How can you guys not see this?

Doing benchmarks this way is way to unpredictable. You have to use the same atmosphere over and over to get a true benchmark.
Thats why they been doing it with a benchmark program for years.

No test are alike ,sometimes not even close.
I proved my point.

I know there are unaccountable variables, but that's part of the point. But you may need to alter your methodology. Do a scene with no variable explosions, or make sure to blow the same things up each time, and replicate your actions. Or do a 10 minute bench to get a larger sample size. Don't stare at the sky on accident.

I was able to reproduce the same fps over and over in BFBC2 when I was testing out the OC on my card by running down and blowing up the same building each time and killing the enemies in the same way.

edit: Btw however you are copying/pasting is misaligned. The sets of fps don't match the summaries at *all*. You last set of fps doesn't show a single sample above 70 but the summary you pasted for it has the overall average at 73.533. Perhaps you are comparing the wrong runs entirely.

 

[HumblePie]

I have to disagree with you on this point. If you have a video card running 80C for 10 hours on air cooling and another running 40C for 10 hours on water cooling, both running the same program, same load, in the same sized room, the ending ambients will not be the same, nor will they reach the same maximum. Saying the 40C watercooled card will heat the room to its maximum ambient sooner than the 80C aircooled card is incorrect. If it was correct, would I be correct saying a -140C card on liquid helium, dry ice, or liquid nitrogen running full load creates the same warmer ambients in a room as does an 80C card on air cooling and reaches that maximum sooner?? ... doesn't make any sense.

Are you disagreeing with physical laws? This isn't something you disagree or argue over period. Thermal heat is being created by a video card chip. You either transfer or translate that heat. Meaning you can turn thermal energy to kinetic for example or you can move the heat. However, you still have the heat to deal with.

Your "perception" using superfluous temps numbers for your hypothetical reasoning does not change this immutable fact. This is not something you argue or disagree with.

Point in fact is that water cooling will not lower ambient temps. To think so is ludicrous. As someone else pointed out, water cooling actually would increase ambient more because of the energy needed by not only fans but a pump as well.

As for the dry ice bit, that is something completely different. You have a energy sink source of something colder than ambient. The dry ice is pulling the thermal energy into itself to bring it up to ambient. That's why using dry ice or liquid nitrogen in a system for cooling won't increase ambient. Now phase change will or refrigerators will as well. Go to howstuffworks.com and read how a refrigerator works for example. It works by pulling the heat out of the air inside the fridge and dumping that heat outside of it. Just a simple transference of heat.

 

[railven]

Here ya go.

Frames, Time (ms), Min, Max, Avg
3397, 60000, 42, 87, 56.617

FPS 55 58 62 49 52 50 51 55 59 66 67 65 63 66 64 63 58 65 64 69 77 87 58 51 51 52 52 53 48 43 50 60 70 71 55 43 42 43 43 47 48 49 45 46 46 45 47 50 52 59 61 61 51 51 60 66 63 65 70 65

Frames, Time (ms), Min, Max, Avg
3557, 60000, 13, 89, 59.283

FPS 57 58 63 59 58 60 64 63 60 62 60 70 66 70 62 57 53 53 54 50 46 51 54 56 54 51 53 53 47 49 56 59 45 46 40 45 45 46 48 55 58 85 76 83 83 84 85 84 85 87 88 89 88 88 84 14 25 25 25 24

Frames, Time (ms), Min, Max, Avg
3512 60000, 15, 241, 73.533

FPS 61 65 58 57 57 62 63 62 60 65 61 53 55 57 52 46 47 53 49 50 57 62 64 53 52 52 54 64 59 60 60 63 60 57 60 67 60 50 49 50 49 50 45 54 51 50 46 49 53 45 41 43 44 45 45 44 43 42 54 55

Did you say margin of error? This is ridiculous, I can't even make it come close to the same on all three runs. Most games don't even do the same thing over and over ,enemies are in different spots, different things blow up, more smoke. How can you guys not see this?

Doing benchmarks this way is way to unpredictable. You have to use the same atmosphere over and over to get a true benchmark.
Thats why they been doing it with a benchmark program for years.

No test are alike ,sometimes not even close.
I proved my point.

In both scenarios you ran it seemed one result was spiked higher than the other two, while these other two were within reasonable distance of each other.

Do you have any kind of background applications running?

Since you proved your point, which do you feel would be a better metric to measure probable game play results? Canned Demos or Game play Demos?

And, lastly, do you acknowledge both side (ATI/nVIDIA) have specifically targeted testing suites and optimized their drivers to produce better results?

If a card could do ~100 FPS on a canned Demo, why can't it do the same in game play? Using you're numbers as the basis for this question. You averaged 80FPS on your canned demo but about 60FPS on your game play demo.

Just curious what you think of that.

 

[busydude]

Frames, Time (ms), Min, Max, Avg 3557, 60000, 13, 89, 59.283 FPS 57 58 63 59 58 60 64 63 60 62 60 70 66 70 62 57 53 53 54 50 46 51 54 56 54 51 53 53 47 49 56 59 45 46 40 45 45 46 48 55 58 85 76 83 83 84 85 84 85 87 88 89 88 88 84 14 25 25 25 24 Frames, Time (ms), Min, Max, Avg 3512 60000, 15, 241, 73.533 FPS 61 65 58 57 57 62 63 62 60 65 61 53 55 57 52 46 47 53 49 50 57 62 64 53 52 52 54 64 59 60 60 63 60 57 60 67 60 50 49 50 49 50 45 54 51 50 46 49 53 45 41 43 44 45 45 44 43 42 54 55 Did you say margin of error? This is ridiculous, I can't even make it come close to the same on all three runs. Most games don't even do the same thing over and over ,enemies are in different spots, different things blow up, more smoke. How can you guys not see this?

Are you serious? Did you just fiddle around with those scores?

Look at the 2nd and 3rd runs carefully, 3557 frames with average of 59.283 and 3512 frames with average of 73.533? How can this be possible?

Now 3557/60 = 59.283 and 3512/60 = 58.53s how did you get an average of 73.533??

So now the averages for 3 runs are 56.617, 59.283 and 58.533 which seem consistent.

 

[railven]

edit: Btw however you are copying/pasting is misaligned. The sets of fps don't match the summaries at *all*. You last set of fps doesn't show a single sample above 70 but the summary you pasted for it has the overall average at 73.533. Perhaps you are comparing the wrong runs entirely.

In his other data set they weren't accurate either. If you look back at it the mins and max's aren't right.

That's it I'm busting out a calculator...

 

[GaiaHunter]

edit: Btw however you are copying/pasting is misaligned. The sets of fps don't match the summaries at *all*. You last set of fps doesn't show a single sample above 70 but the summary you pasted for it has the overall average at 73.533. Perhaps you are comparing the wrong runs entirely.

Yeah. His first 2 BFBC2 don't add up either.

First adds to 5571 frames over 60s or 92.85 average vs 6050 frames reported and that would sum to 100.833 avg (which is the reported).

Seconds adds to 6049 frames over 60s or 100.81 average vs 5972 frames reported or 99.53 avg when reported is 92.867.

It is a mess.

And his FC2 seem even more problematic. And of course nowhere 80fps of the canned.

 

[railven]

Are you serious? Did you just fiddle around with those scores?

Look at the 2nd and 3rd runs carefully, 3557 frames with average of 59.283 and 3512 frames with average of 73.533? How can this be possible?

Now 3557/60 = 59.283 and 3512/60 = 58.53s how did you get an average of 73.533??

So now the averages for 3 runs are 56.617, 59.283 and 58.533 which seem consistent.

Haha, there you go. Looking at the third data set, there are more <=65's than >70's just using that as a base there is no way he got 73.533 FPS average.

How misleading. There is no way he copied and pasted that wrong, he must have deliberately changed the values.

 

[MrK6]

I have to disagree with you on this point. If you have a video card running 80C for 10 hours on air cooling and another running 40C for 10 hours on water cooling, both running the same program, same load, in the same sized room, the ending ambients will not be the same, nor will they reach the same maximum. Saying the 40C watercooled card will heat the room to its maximum ambient sooner than the 80C aircooled card is incorrect. If it was correct, would I be correct saying a -140C card on liquid helium, dry ice, or liquid nitrogen running full load creates the same warmer ambients in a room as does an 80C card on air cooling and reaches that maximum sooner?? ... doesn't make any sense.

Well, heat's a quantity, and however much power you're using will pretty much all be turned into heat and then be transferred to the room. However, you may notice cooler ambient temperatures with a card on water because chips running colder consume less power. So less heat is introduced to the environment because the card is consuming less power at 40C than at 80C, so what you're seeing is an indirect effect on heat output but running a part at colder temperatures. If you add in sub-ambient cooling, it complicates it further because you're introducing a "cold" sink.

Are you serious? Did you just fiddle around with those scores?

Look at the 2nd and 3rd runs carefully, 3557 frames with average of 59.283 and 3512 frames with average of 73.533? How can this be possible?

Now 3557/60 = 59.283 and 3512/60 = 58.53s how did you get an average of 73.533??

So now the averages for 3 runs are 56.617, 59.283 and 58.533 which seem consistent.

:hmm:

 

[railven]

I used all the values on his FarCry 2 #3 run and got the following:

Total Frames == 3234
Total Counts == 60
Total Frames Average == 53.9

Wow, Happy would blatantly lie to prove his point.

Added the values for run #1 and #2 and they are exactly what he posted.

 

[Keysplayr]

Just wondering if anyone paid attention or even acknowledged ViRGE's post a few back.
It was actually interesting and something I did not know.

 

[EightySix Four]

I have to disagree with you on this point. If you have a video card running 80C for 10 hours on air cooling and another running 40C for 10 hours on water cooling, both running the same program, same load, in the same sized room, the ending ambients will not be the same, nor will they reach the same maximum. Saying the 40C watercooled card will heat the room to its maximum ambient sooner than the 80C aircooled card is incorrect. If it was correct, would I be correct saying a -140C card on liquid helium, dry ice, or liquid nitrogen running full load creates the same warmer ambients in a room as does an 80C card on air cooling and reaches that maximum sooner?? ... doesn't make any sense.

Think about the heat output as energy output instead of temperature, both cards are releasing 140 watts (random out of my ass number), one system is just getting that energy away from the card and into the room better, meaning the temperature of the card is lower. In both situations, the same amount of energy is being released in to the room.

Just wondering if anyone paid attention or even acknowledged ViRGE's post a few back.
It was actually interesting and something I did not know.

It's true and useful to know, but without the actually specifics of the process these chips are developed on, we have literally no way of knowing if this is the difference of .5w, or 50w.

 

[GaiaHunter]

I used all the values on his FarCry 2 #3 run and got the following:

Total Frames == 3234
Total Counts == 60
Total Frames Average == 53.9

Wow, Happy would blatantly lie to prove his point.

Added the values for run #1 and #2 and they are exactly what he posted.

Nah.

It has to be a mistake or problem somewhere.

No one would be so inept to be lying while providing the numbers (and numbers that jump out) to prove him wrong.

 

[GaiaHunter]

And why are you still discussing the temperature of the room?

It is simple - in the case of LN2 and such, the room temperature doesn't increase because LN2 suffers a physical change of state - LN2 vaporizes.

That requires energy to happen.

Air doesn't change state (in this case of cooling PC parts), so it increases temperature as it absorbs the heat. Same as water (in this case of cooling PC parts) that then is cooled by the air dumping its heat in surrounds which increases temperature of room.

http://en.wikipedia.org/wiki/Change_of_state

http://en.wikipedia.org/wiki/Phase_&#37;28matter)

And so on.

 

[railven]

Nah.

It has to be a mistake or problem somewhere.

No one would be so inept to be lying while providing the numbers (and numbers that jump out) to prove him wrong.

That is very discouraging. Happy get's flagged for derailing threads in another thread and here he is now posting inaccurate results to further support his claim.

And now the topic is heading back towards temperatures?

 

[luv2increase]

And now the topic is heading back towards temperatures?

Sadly, it is not a real concern as people who've never owned the cards make it out to be. It is pitiful indeed to see people complain about the temperature in every sarcastic manner possible yet have never owned the cards themselves to experience the "fact" it is a non-issue, even in multi-GPU configs with Fermi.

I really pity them