My benches so far. Adding GTX480 SLI benches.

Page 13 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
You would have to benchmark without it, or else you would rarely see any differences between any of the cards. Benchmarking with it off is totally acceptable and supported by me. I don't think it should be done any other way. Just wanted to clarify... But for actual game-play, I won't ever run without it.
If you benchmarked it with it on, the difference would be in the average FPS and the minimums, but not as glaringly obvious or different than had it been off. Of course, if you're running a game where all cards tested can consistently output 60FPS+, well, yeah, tough to find a difference there :p.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
Ah,

nKey's..my friend...

What's this? Is this you making me "pay the price"?
Joking, or not, we can do without these little stabs and have an adult conversation.
Maybe next time you'll take my posts as real without Apoppin's back up.
And yes, we know. The card is hotter than it's competition.
In the end though, I'm willing to bet that anyone interested in these level of cards is looking for performance first and foremost, features next, then heat/power. I'm sure they are all considered, but something tells me performance takes the crown here when shopping for a high end graphics cards. Why do you think they call them "high end"?
 

Magusigne

Golden Member
Nov 21, 2007
1,550
0
76
What's this? Is this you making me "pay the price"?
Joking, or not, we can do without these little stabs and have an adult conversation.
Maybe next time you'll take my posts as real without Apoppin's back up.
And yes, we know. The card is hotter than it's competition.
In the end though, I'm willing to bet that anyone interested in these level of cards is looking for performance first and foremost, features next, then heat/power. I'm sure they are all considered, but something tells me performance takes the crown here when shopping for a high end graphics cards. Why do you think they call them "high end"?

Derek has told me to play nice;) Gotta put on my kid gloves.

At any rate. Do you or Apoppin have access to a waterblock with these chips? I'd like to see them underwater and perform. The temperature is definately going to prevent you guys from overclocking them too much.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Derek has told me to play nice;) Gotta put on my kid gloves.

At any rate. Do you or Apoppin have access to a waterblock with these chips? I'd like to see them underwater and perform. The temperature is definately going to prevent you guys from overclocking them too much.

no. Not yet.

i haven't personally seen temps over 100C; Max is 105C when it is supposed to throttle
- but then i have superb case airflow with five case fans (120 - 140 mm)

max stable OC i got with GTX 480 was 825/1100 (i might have pushed a bit higher but didn't want the fan at 100%) and it is a *performance beast* at that OC - ask me about the Radeon HD 5870 at 975/1300

:)
 
Last edited:

Magusigne

Golden Member
Nov 21, 2007
1,550
0
76
Yikes. I don't think I'm cool enough (hahaha pardon the pun) to get a GTX 480 as much as I might like.

If your case is pretty well chilled my case is going to end up seriously cooking.

Nice benches appopin.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Yikes. I don't think I'm cool enough (hahaha pardon the pun) to get a GTX 480 as much as I might like.

If your case is pretty well chilled my case is going to end up seriously cooking.

Nice benches appopin.

Thanks. Yes, i finally got a decent midtower Gladiator 600 case and stuck 3 more big fans in; since my my PSU acts like an exhaust fan also, i have 3 intake and 3 exhaust fans total.

If the inside case temps only raise by 10 or so degrees F over ambient temps, i consider that pretty good.



It's supposed to throttle at 105C???!
What happened to 125?
Well the spec says 105C as "GPU Thermal Threshold" and it adds (from the Reviewer's Guide), "if for any reason the GPU exceeds this temperature, the clock speed will automatically be dialed down"
 

evolucion8

Platinum Member
Jun 17, 2005
2,867
3
81
The card is hotter than it's competition.
In the end though, I'm willing to bet that anyone interested in these level of cards is looking for performance first and foremost, features next, then heat/power. I'm sure they are all considered, but something tells me performance takes the crown here when shopping for a high end graphics cards. Why do you think they call them "high end"?

Hence, the HD 5970 is a better buy. Lower power consumption, better performance and lower temperatures. The same that happened to the HD 2900XT vs 8800GTX and guess who won...
 

MarcVenice

Moderator Emeritus <br>
Apr 2, 2007
5,664
0
0
So keys, are you going to run more tests on your OC instead of only the Stalker-benchmark?

Btw, I managed to draw 750W out of the wall with two GTX 480's in SLI. Ingame. In furmark, powerdraw was 800W. That's 300W more then with a single GTX 480 :p The back of the case was like a hairdryer :D
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
300W more out of the wall does support 250W TDP for the card. Lop 20&#37; off 300 for PSU and wiring inefficiency and you're there. That's not bad at all, I'd have expected closer to a 400 watt increase with a second card. Are you sure it's not overheating and throttling?

I do wonder how many 20C->100C cycles the PCB can take before expiring even if the GPU survives. Also how long until dust and particle buildup pushes normal operating temperature over 105C. Either of you care to take up smoking long enough to see if this happens?
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
So keys, are you going to run more tests on your OC instead of only the Stalker-benchmark?

Btw, I managed to draw 750W out of the wall with two GTX 480's in SLI. Ingame. In furmark, powerdraw was 800W. That's 300W more then with a single GTX 480 :p The back of the case was like a hairdryer :D

When I get some time.
And congratulations on your power draw numbers. Really tugging on some power there. So how was the performance and scaling with SLI?
 

MarcVenice

Moderator Emeritus <br>
Apr 2, 2007
5,664
0
0
When I get some time.
And congratulations on your power draw numbers. Really tugging on some power there. So how was the performance and scaling with SLI?

Didn't compare all games yet, but in Crysis Warhead, Dawn of War II: Chaos Rising and Far Cry 2, minima doubled, avg sometimes did too, but not always because of a slight cpu-bottleneck. I expect scaling to be 'superb'.

Powerdraw is crazy :p At first, during some short benchmarks the noise was acceptable, but depending on the game and how long you play it, the videocards really do heat up, in a HAF 932, which has great airflow.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
Didn't compare all games yet, but in Crysis Warhead, Dawn of War II: Chaos Rising and Far Cry 2, minima doubled, avg sometimes did too, but not always because of a slight cpu-bottleneck. I expect scaling to be 'superb'.

Powerdraw is crazy :p At first, during some short benchmarks the noise was acceptable, but depending on the game and how long you play it, the videocards really do heat up, in a HAF 932, which has great airflow.

Cool, let us know when you have some printed results. I'd like to see them. Scaling is supposed to be improved in the GF100 series.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
Even if he did it, you'd just dispute and argue those results too so it's a lose-lose situation for him no matter what. He loses lots of time finding the max stable o/c for each card and then rerunning benchmarks, and he loses with people like you who, regardless of the results, won't believe key's methodology anyhow.

I agree with you 100%. That is why, after voicing my opinion (somewhere near the beginning of this thread, or maybe apoppin's thread) about the validity and purpose of these tests, I stopped. I had my say, it was countered, and that's the end of it. No point in going on ad nausium.
 

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
Are u going to add any gtx 470 benches?(coz i'm buying one recently :))Also can u shed some light on the Zotac brand,as in my place gtx 470 is only available from zotac.Thanx bye.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
Are u going to add any gtx 470 benches?(coz i'm buying one recently :))Also can u shed some light on the Zotac brand,as in my place gtx 470 is only available from zotac.Thanx bye.

I would if I had any.
Zotac is fine, just pay attention to whatever warranty policy they have. It may be a year, 2 years, 3 years, I don't know. But please pay close attention to the warranty policies before you buy.
 

Lonyo

Lifer
Aug 10, 2002
21,939
6
81
My pleasure. It is sick, isn't it?

Greater than 100% scaling on minimum frame rates is indeed very sick.
Specifically Stalker COP "Day", where it goes from 25fps minimum with a single card, to 54fps minimum with SLI.

Question: What is your methodology? Do you do multiple runs and take the average, multiple and discard anomalous, multiple and check whether they are within a certain % of each other?
 

Jaydip

Diamond Member
Mar 29, 2010
3,691
21
81
I would if I had any.
Zotac is fine, just pay attention to whatever warranty policy they have. It may be a year, 2 years, 3 years, I don't know. But please pay close attention to the warranty policies before you buy.

They are offering 3 years warranty.I think i can safely assume it will be slower then a 5870 by 10% (on average) and will be faster than a 5850 by 10%(on average).Also can u tell if a corsair tx 750 would suffice for gtx 470?.Thanx for your time.bye.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
Greater than 100% scaling on minimum frame rates is indeed very sick.
Specifically Stalker COP "Day", where it goes from 25fps minimum with a single card, to 54fps minimum with SLI.

Question: What is your methodology? Do you do multiple runs and take the average, multiple and discard anomalous, multiple and check whether they are within a certain % of each other?

I do 4 runs of each benchmark. I discard the first run as this run offers the most drive thrashing loading textures. I then take the average of the remaining 3 runs Min, Avg and Max.

My methodology has always been this way. For both NV and ATI benches.