Thursday, October 1st 2009

ASUS HD 5870 Overclocks to 1035/1290 MHz on Air, Aces 3DMark Vantage in CrossFireX

by
btarunr
 Discuss (57 Comments)
Here is what four AMD Cypress GPUs can achieve with some careful overclocking, without needing any third-party cooling. Renowned overclocker Kinc sent us details of his latest achievement using four ASUS Radeon HD 5870 1 GB cards installed in a 4-way CrossFireX setup, all overclocked, and cooled by AMD's reference cooler, taking a shot at 3DMark Vantage (Extreme Preset). The four cards returned a score of X26,332 points, with an average frame-rate of 79.49 fps in GT1, and 74.83 fps in GT2.

To begin with the cards were overclocked to 1035/1290 MHz, up from reference speeds of 850/1200 MHz (core/memory). This was supported by raising the vGPU to 1.330V using GPUTool, from 1.015V. The platform to drive this feat comprised of an Intel Core i7 965 XE processor, cooled by Intel's reference (boxed) cooler, clocked at 4257 MHz. To seat them all was an ASUS P6T7 WS SuperComputer motherboard. The feat serves as a prelude to what the future holds in two "Hemlock" accelerators, which make use of two Radeon HD 5870 GPUs each.

Related News

Add your own comment

57 Comments on ASUS HD 5870 Overclocks to 1035/1290 MHz on Air, Aces 3DMark Vantage in CrossFireX

#26
AsRock
TPU addict
1.330V WOW thats way lower than the 4890's NICE!..
Posted on Reply
#27
aCid888*
BingeYeah, and it still did the same as my OCed 295. The temps still must have sucked. You think I don't see it ran the benchmark? :wtf:
Links to your single 295 doing X26k in Vantage?? :rolleyes:





These cards are monsters, I cant wait to see how they do with better drivers as I'm sure they dont scale well with the ones they are using at the moment.

A better clocking D0 would work wonders too. :toast:
Posted on Reply
#28
Unregistered
Just for fun, is there anyone out there who has access to four of these and could do a full review?

It would give a good indication of what a pair of X2's could do. Plus it would be plain old cool to see.:)
#29
Binge
Overclocking Surrealism
aCid888*Links to your single 295 doing X26k in Vantage?? :rolleyes:





These cards are monsters, I cant wait to see how they do with better drivers as I'm sure they dont scale well with the ones they are using at the moment.

A better clocking D0 would work wonders too. :toast:
That's Quadfire sir... I was talking about the P numbers the single cards got OCed at that ammount.
Posted on Reply
#30
SonDa5
15th WarlockYou mean this guy invested over $1,600 in video cards and couldn't afford at least $80 for a decent aftermarket CPU cooler? :p

I mean, if you plan to buy a Ferrari, you may as well have some extra change for the gas, don't you agree? :laugh:

Just kidding, I'm so envious, I'm pretty sure he'll come up with even more impressive benchmarks after getting some decent cooling for his killer setup :rockout:
Yeah. Looks like he is just familiarizing himself with his new toys and throwing some base benchmarks out there. Kind of like a lion roaring.

I'm sure the HD5870 is going to destroy all the benchmark records soon.
Posted on Reply
#31
1Kurgan1
The Knife in your Back
BingeYeah, and it still did the same as my OCed 295. The temps still must have sucked. You think I don't see it ran the benchmark? :wtf:
I was joking mildly, if you couldnt tell by ":p", but I will explain, when benchmarking who cares what your temps are, your going for top numbers, I highly doubt he plans on running this 24/7 so who cares.

And beyond that, if your single 295 scores that much in extreme, then looks like your in the top 5 for world records according to the futuremark site, you might want to register that over there. ;) I'm thinking you mean performance and didn't see that is what this guy was running here.

**EDIT**

I see you meant you beat a single 5870 with a dual GPU card, not a big shock there :banghead:
Posted on Reply
#32
PP Mguire
15th WarlockYou mean this guy invested over $1,600 in video cards and couldn't afford at least $80 for a decent aftermarket CPU cooler? :p

I mean, if you plan to buy a Ferrari, you may as well have some extra change for the gas, don't you agree? :laugh:

Just kidding, I'm so envious, I'm pretty sure he'll come up with even more impressive benchmarks after getting some decent cooling for his killer setup :rockout:
He didnt spend a dime on any of this. And what they didnt mention, is the stock cooler prolly had dry ice on it.
1Kurgan1I was joking mildly, if you couldnt tell by ":p", but I will explain, when benchmarking who cares what your temps are, your going for top numbers, I highly doubt he plans on running this 24/7 so who cares.

And beyond that, if your single 295 scores that much in extreme, then looks like your in the top 5 for world records according to the futuremark site, you might want to register that over there. ;) I'm thinking you mean performance and didn't see that is what this guy was running here.

**EDIT**

I see you meant you beat a single 5870 with a dual GPU card, not a big shock there :banghead:
Exactly what i was thinking. Now when 2 GTX295s can beat 4 of these in Xtreme then ill be impressed.
Posted on Reply
#33
SonDa5
BingeThat's Quadfire sir... I was talking about the P numbers the single cards got OCed at that ammount.
There is a huge difference between "X" and "p".

"X" scoring is tougher and it is superior.
Posted on Reply
#34
Binge
Overclocking Surrealism
1Kurgan1I see you meant you beat a single 5870 with a dual GPU card, not a big shock there :banghead:
So you're saying nobody says the 5870 > GTX295 :rolleyes: I can tell you're joking mildly again. It's a serious comment but there's something someone could laugh at. People care about heat too. Get it? I care about heat a lot. Understand? Sheesh... Grab a beer, k? :toast:
Posted on Reply
#35
SonDa5
BingePeople care about heat too. Get it? I care about heat a lot. Understand? Sheesh... Grab a beer, k? :toast:
The truth is that without real system temp information it's impossible for anyone to guess how hot his GPUs were.
Posted on Reply
#36
Binge
Overclocking Surrealism
SonDa5The truth is that without real system temp information it's impossible for anyone to guess how hot his GPUs were.
Yeah that's the case, and it's a total bum scenario.
Posted on Reply
#38
icon1
X27,253... extreme :cool:
Posted on Reply
#39
Frizz
Can't wait for Nvidia to release their GT300, when they do, its price wars again. I always stick with the bang for buck side and usually that's ATI. Can't wait to get my hands on these babies.
Posted on Reply
#40
1Kurgan1
The Knife in your Back
BingeSo you're saying nobody says the 5870 > GTX295 :rolleyes: I can tell you're joking mildly again. It's a serious comment but there's something someone could laugh at. People care about heat too. Get it? I care about heat a lot. Understand? Sheesh... Grab a beer, k? :toast:
The reason you got the faces is because you seemed to take offense to my perviously light hearted statement. Either way when it comes to synthetics I would say the 295 is going to beat out 1 5870, dual gpu's scale better in synthetics. And like I said with heat, I highly doubt this is his 24/7 clocks. This guy is known for going for world records, you don't ask the guys running DICE what their temps are, because it just don't matter how cold or warm it is, if it sets a record it sets a record, even if it's on fire.

I understand about caring about heat, trust me I have a furnace of a 4870x2 and when I touch the backplate on it, it could literally burn me if I left my hand on it. :laugh: But yeah this guy on't be running this that aggressive 24/7 as it just isnt needed.
Posted on Reply
#41
jamesrt2004
wonder if he got betetr score with v-ram a bit lower, as we know if memory clocked too high it will do some stuff to slow it down a tad etc.. wonder how high it can go before then?
Posted on Reply
#42
icon1
Definitely should clock higher with water blocks installed in it.
Just waiting for GT300 cards to arrive before making any further upgrade. Nice scores indeed! :D
Posted on Reply
#43
WarEagleAU
Bird of Prey
Man this is bad ass. Lets see what Nvidia has to say.
Posted on Reply
#44
SystemViper
WOW, it looks like you need 4 gpu to just play in this sandbox, got to love tech, but man the costs, lol......

I can't wait for winter to kick in and break out the igloo.... then I think we'll hear Michael Buffer in the distance "Let the games begin"

got to love the holiday season, new chips, new motherboards and new GPU's...... :toast:
Posted on Reply
#45
aCid888*
I'm busy collecting parts for my winter bench session outside in the really warm -30'C (!) Canadian air. :roll:

I dont have 4 5870's but I do own 3 4870's. :)
Posted on Reply
#46
Imsochobo
aCid888*I'm busy collecting parts for my winter bench session outside in the really warm -30'C (!) Canadian air. :roll:

I dont have 4 5870's but I do own 3 4870's. :)
ill get a 5870, watercooling it, and enjoying the same cold air you got in norway :D

Ill just take a long watertube and place the damn radiator outside :D

I ended up with temps at -4 C(you really cant shut it down cause it'll freeze!)
Posted on Reply
#47
LAN_deRf_HA
Can someone test a single 5870 at those speeds against a 295? That overclock should be enough to beat it, and maybe make it on par with a 380 or w/e nvidia is going to call it. Considering a 280 performed less than the GX2.
Posted on Reply
#48
PP Mguire
...What are you saying? :laugh: :wtf:

Ok so your saying overclock this to be on par with a 380? That would put the 380s performance incredibly high.

And the 9800GX2 does not beat a 280.
Posted on Reply
#49
LAN_deRf_HA
When the 280 GTX first came out it did not outright beat a 9800 GX2, it traded blows with it. Just like how the 5870 trades blows with the 4870 X2. Doubling or slightly more than doubling the shader count is not enough to completely beat the dual gpu cards of the prior gen. Considering the 295 is only 15% above a 5870, I'd say a 5870 clocked to 1035mhz (21% oc) could beat a 295. Thus matching or coming close to a 380 (what I assume it will be called), this with an overclock it can do on stock cooling. This is interesting to me because a lot of people are acting like the 380 is going to be some insane step forward, leaving the 5870 completely in the dust. I don't think that will really be the case based on the numbers.

It would however be nice for someone to take a single 5870, bump it to 1035 and bench it against a 295 GTX to confirm it scales that well.
Posted on Reply
#50
Bo_Fox
btarunrHere is what four AMD Cypress GPUs can achieve with some careful overclocking, without needing any third-party cooling. Renowned overclocker Kinc sent us details of his latest achievement using four ASUS Radeon HD 5870 1 GB cards installed in a 4-way CrossFireX setup, all overclocked, and cooled by AMD's reference cooler, taking a shot at 3DMark Vantage (Extreme Preset). The four cards returned a score of X26,332 points, with an average frame-rate of 79.49 fps in GT1, and 74.83 fps in GT2.

To begin with the cards were overclocked to 1035/1290 MHz, up from reference speeds of 850/1200 MHz (core/memory). This was supported by raising the vGPU to 1.330V using GPUTool, from 1.015V. The platform to drive this feat comprised of an Intel Core i7 965 XE processor, cooled by Intel's reference (boxed) cooler, clocked at 4257 MHz. To seat them all was an ASUS P6T7 WS SuperComputer motherboard. The feat serves as a prelude to what the future holds in two "Hemlock" accelerators, which make use of two Radeon HD 5870 GPUs each.

www.techpowerup.com/img/09-10-01/85a_thm.png
Wow, what an impressive "show-off" of what a stock-cooled setup could do!!!

Well, few questions..

The stock voltage is supposed to be just a little over 1.1v, right? 1.015v is much lower than what I've heard!

What were the max temps of the GPU's? I'll take the liberty of assuming that all of the fans ran at 100% speed.

Last question: Did he have to try more than 4 cards to "cherry-pick" the good overclockers? If he never got any more than 4 cards to begin with, and just overclocked all of them to 1+ GHz speeds using stock coolers, then I surely would be enlightened with those Asus cards! :D
Posted on Reply
Add your own comment
May 2nd, 2024 22:35 EDT change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts