*** Official Radeon HD5800 Series Overclocking Thread ***

[SonDa5]

Not sure if I have posted my HD5850 Vantage work.






Orb details with PR:

http://service.futuremark.com/result...eResultType=19

[Bojamijams]

Whats weird is that I pass vantage fine at 1300 mem but running Wow, I'll get graphical corruption eventually.. at 2x anti aliasing and full shadows and full everything at 1920x1200.. the driver recorvers but still.. 10 second pause

[Mungri]

Good to know.

Between single HD5850 and HD5770 Xfire which do you think is better? Why?

I cant really say because I havnt used the 5850's.

I went with the 5770's because rewiews showed two of them being as good as a single 5870 while costing less.

If you get just one 5850 or 5870, you can still add a second later and get uber high end performance, two 5770's on the other hand is the price / performance choice, but I'll say that for most people, a single 5850 or 5870 would be a better choice.

[ajaidev]

If this is true then i am not impressed. My friend on a 5870 @ 1.1ghz has scores very close to this and his i7 920 is around 3.7-3.8Ghz....

[spajdr]

ajaidev : well cores are only 725Mhz, at that resolution you are badly limited by CPU, even if its running at 4Ghz, so i dont think that score is that bad

[jaredpace]

yah it's exactly the speed of 5850 x-fire, maybe a little faster. I bet if you had this a 1.1ghz it'd score close to 30k gpu



edit: this is from Bodar @ guru3d.
It appears the ABT benchmarks are done on a dual 1440SP card, because here is the comparison with dual 1600SP at the same clocks + same cpu.
(Look at the 3DV CPU score.)

Bodar's Settings


ABT ES 5950


Bodar's 5870 Crossfire


ABT ES 5950


Bodar's 5870 Crossfire


ABT ES 5950


Bodar's 5870 Crossfire


http://forums.guru3d.com/showpost.ph...1&postcount=19

Summary:

3Dmark Vantage (P score - GPU):

5850 CF 1440sp 725/1000 - P21308 / 20734 (87%)
5950 ES 1440sp 725/1000 - P21940 / 21151 (89%)
5870 CF 1600sp 725/1000 - P22891 / 22342 (95%)
5870 CF 1600sp 850/1200 - P23308 / 23657 (100%)

Crysis Warhead (min - avg - max):

5950 ES 1440sp 725/1000 - 20.4, 34.2, 42.2
5870 CF 1600sp 725/1000 - 24.6, 37.4, 46.2

Stalker Clear Sky (average frames):

5950 ES 1440sp 725/1000 - 45.7, 43.8, 47.7, 35.6
5870 CF 1600sp 725/1000 - 49.8, 49.3, 54.7, 38.1

[jfer]

Hi! Are these temperatures on Voltage regulators scary?

[zalbard]

Definitely higher than average. But should be safe. Don't worry about them. If that's the highest they go up to of course.

[SonDa5]

Hi! Are these temperatures on Voltage regulators scary?

Very scary. With no load looks like it's going to melt under load. I'd go back to stock cooler till the problem is fixed.


Also nice 3dMark06 score. I'm just a little higher. (with Q9550@4.3ghz)
Your CPU has an advantage and I'm sure with some over clocking your can hit 26,000+

[SonDa5]

If this is true then i am not impressed. My friend on a 5870 @ 1.1ghz has scores very close to this and his i7 920 is around 3.7-3.8Ghz....

That doesn't look promising for i7.

Maby the newer i9 builds can get it to go.

[zalbard]

Very scary. With no load looks like it's going to melt under load. I'd go back to stock cooler till the problem is fixed.

They are designed to work at up to 110 C afaik.
Temps are high, but not horrible. I'd get worried if they started going over 80 C, though.
And he isn't the only one.
See here. Pretty close. Just a random link that took 30 seconds to find.
But if it's going over 80 C under load, then I agree, definitely, look into the problem, it's not normal IMO.

[jfer]

Very scary. With no load looks like it's going to melt under load. I'd go back to stock cooler till the problem is fixed.


Also nice 3dMark06 score. I'm just a little higher. (with Q9550@4.3ghz)
Your CPU has an advantage and I'm sure with some over clocking your can hit 26,000+

Thats the max Temps during 3 Loops of 3dmark06. At idle they stay between 38~42ºC.

[zalbard]

Thats the max Temps during 3 Loops of 3dmark06. At idle they stay between 38~42ºC.

That's okay then!
Not going to kill your card for sure. Just a bit warmer than for most folks here. Maybe disassemble the cooler to check if everything is fine if temps start rising.

[SonDa5]

Thats the max Temps during 3 Loops of 3dmark06. At idle they stay between 38~42ºC.

Your GPU-Z screen shot shows it ideling with o% load. Your GPU core isn't that hot but your other temps look hot.

[Mungri]

I got my 5770s stable at 1000 Mhz, needed the fan to be put very loud, and 1.225v:



They definately wont go up to 1050 Mhz though, even with 1.35v and the fan set to 100%, everything artifacts.

By stable, I mean that I've been playing Battleforge for over an hour without any problems, and the temps are below 90 degrees with a noisy fan. I need to wait for some compatible VRM coolers before I can start modifying them.

At 100% default fan and clock settings, the first card was going up to 97-100+ degrees in battleforge.

They've been fine for over 3 hours of Battleforge now and remained under 90 degrees at 88 max

[Gamer]

I've bought this Riser to work with 4 cards, but it seems it doesn't work.
I guess I bought the wrong one ?
http://cgi.ebay.co.uk/Flexible-PCI-E...InterfaceCards

Can someone tell for sure ?

thanks.
EVGA Classified 759.

[Rattle]

wow 5000 series are so sexy

[gabe]

in response to http://www.xtremesystems.org/forums/...&postcount=507

Definitely higher than average. But should be safe. Don't worry about them. If that's the highest they go up to of course.

While I do agree with your statement from a strict technical standpoint, I can't help but feeling quite frustrated by the "don't worry" advice. As manufacturers (and I am sure other colleagues might agree with me if they read this), we spend countless hours to fine-tune every aspect of our cooling systems to lower all the critical temps to a minimum. There was a long discussion earlier in this thread where ppl agonized about the shape of the water-block flow path to see if it was optimized to cool the VRM's..

jfer didn't specifiy what type of waterblock he is using, full cover, or gpu only. If it is a full-cover block, these temps are not worth the money he spent on the product (or RMA is in order). If he is using a gpu only cooler and individual sinks, then some ventilation on the sinks wouldn't hurt.

Also, I don't know if ATI has a temperature shutdown associated with the VRM's, but I found out working on the dual board GTX295 that nVidia has. The Board would lock up in high voltage high overclocks pushing the VRM's beyond 90 C.

[ajaidev]

yah it's exactly the speed of 5850 x-fire, maybe a little faster. I bet if you had this a 1.1ghz it'd score close to 30k gpu

edit: this is from Bodar @ guru3d.
It appears the ABT benchmarks are done on a dual 1440SP card, because here is the comparison with dual 1600SP at the same clocks + same cpu.
(Look at the 3DV CPU score.)

Summary:

3Dmark Vantage (P score - GPU):

5850 CF 1440sp 725/1000 - P21308 / 20734 (87%)
5950 ES 1440sp 725/1000 - P21940 / 21151 (89%)
5870 CF 1600sp 725/1000 - P22891 / 22342 (95%)
5870 CF 1600sp 850/1200 - P23308 / 23657 (100%)

Crysis Warhead (min - avg - max):

5950 ES 1440sp 725/1000 - 20.4, 34.2, 42.2
5870 CF 1600sp 725/1000 - 24.6, 37.4, 46.2

Stalker Clear Sky (average frames):

5950 ES 1440sp 725/1000 - 45.7, 43.8, 47.7, 35.6
5870 CF 1600sp 725/1000 - 49.8, 49.3, 54.7, 38.1

I think that is correct 100%, this may as well be the 1440 * 2 card. But i hope AMD calls it 5950 and not 5970. Well the clock speed and memory speed make sense and so do the performance and price "This is why its suppose to be so close to the 5870 price"

If they call it 5970 that would be wrong representation but i guess they can call it what ever the hell they feel like. I would have like a shorter PCB since the 4 * 5850's we have for milkyway@home is just rite to fit in a normal case without any mods.

In the end i would buy either this or the GF100 depending on the performance and how much they cost. I for one would have loved a 5890 with clock speeds around 1Ghz and special vaporX cooler. Would have not mind spending some extra bucks on that but it should be priced under the GF100....

[Gaul]

ABT ES 5950

i'll compare with 5870 def + q66 @ 4 gH

[zalbard]

in response to http://www.xtremesystems.org/forums/...&postcount=507



While I do agree with your statement from a strict technical standpoint, I can't help but feeling quite frustrated by the "don't worry" advice. As manufacturers (and I am sure other colleagues might agree with me if they read this), we spend countless hours to fine-tune every aspect of our cooling systems to lower all the critical temps to a minimum. There was a long discussion earlier in this thread where ppl agonized about the shape of the water-block flow path to see if it was optimized to cool the VRM's..

jfer didn't specifiy what type of waterblock he is using, full cover, or gpu only. If it is a full-cover block, these temps are not worth the money he spent on the product (or RMA is in order). If he is using a gpu only cooler and individual sinks, then some ventilation on the sinks wouldn't hurt.

Also, I don't know if ATI has a temperature shutdown associated with the VRM's, but I found out working on the dual board GTX295 that nVidia has. The Board would lock up in high voltage high overclocks pushing the VRM's beyond 90 C.

Alright, my bad.
I read the post, but I missed the post's title, which was "Watercooled 5850@1000/1300". So I was assuming he was on air. Hence the misunderstanding.
Since he is using liquid cooling, I have to agree with you completely.
Just one thing to add.
This post: http://www.xtremesystems.org/forums/...&postcount=398
Perhaps jfer can tell us if it's under water or not. I assume it is air judging the fan speed reported by GPU-Z. If it's the same water cooling solution as the one he used for the last run, then something is going wrong with the waterblock, or with the card, since VRM temperatures have significantly increased since then.
And one more thing, I wouldn't be surprised if this were his VRM temps for GPU-only waterblock. VRMs can easily get hot as hell. Perhaps he might want to have a look at the special VRM solution made by Thermalright.

[jfer]

Im having my 5850 with a D-Tek Fusion(not lapped), BGA ram sincks, and 2Revoltec 50mm on the 4regulators at 4100rpm(very low noise btw).
Here some pics:

[R3DCrocker1]

Did some 01 this weekend. Had some strange issues. Card wouldnt clock to 1120 on dice on the asus bios and topped out at 920 on the MSI one.
Doesnt matter though. i got a strong nature card! over 6ghz on the CPU wil give me gold!

[Mungri]

If I remover the 5770 stock cooler for an aftermarket one, where abouts, other then the GPU and Ran chips do I need to attack heatsinks to?



I have two packs of 10 mofset heatsinks ordered that will fit on the 9 black chips next to the Yageo things, so i need heatsinks anywhere else?

[Icebob]

Here's my vantage work on a P55 platform Trifire 5870

[IMG][/IMG]