[H] users 7950 Overclock Results

Sapphire HD7950 950mhz edition here.



Voltage goes to 1.38v on core.
Clocks are unlocked on core.
Switch for dual BIOS.
Built on reference HD7970 pcb.
Waiting for water block before pushing hard.
Stock heat sink clears benchmarks and stable gaming up to 1250mhz core at 1.25v.


oc in progress.
 
grambo said:
Personally, I would not care about Furmark temps. Furmark is useful for testing OC stability/artifacting, but it is not representative of a gaming load. I would play BF3 with vsync off (to load the card more heavily) for 2-3 hours on 64 player maps with max detail and if temps are ok, call it stable.
Click to expand...

thanks, ill try that
 
just thought i would post to keep track and for feedback

Running a regular Powercolor 7950 800MHZ 3GB (cheapest one i found)

Core - 1275mhz @ 1225mv
Mem - 1600mhz (havent really touched this much)

Load temp sits at 41C WCed and im using Afterburner for OCing

how am i doing? :D
 
Krayzie135 said:
just thought i would post to keep track and for feedback

Running a regular Powercolor 7950 800MHZ 3GB (cheapest one i found)

Core - 1275mhz @ 1225mv
Mem - 1600mhz (havent really touched this much)

Load temp sits at 41C WCed and im using Afterburner for OCing

how am i doing? :D
Click to expand...

Wow. If you are stable for gaming at 1275 Mhz with a ref HD 7950 card you truly have a golden sample. enjoy it. :)
 
Krayzie135 said:
just thought i would post to keep track and for feedback

Running a regular Powercolor 7950 800MHZ 3GB (cheapest one i found)

Core - 1275mhz @ 1225mv
Mem - 1600mhz (havent really touched this much)

Load temp sits at 41C WCed and im using Afterburner for OCing

how am i doing? :D
Click to expand...

Doing great if it is stable.

Looks like a very good clock speed. Voltage seems a little high though.

Have you done any stability testing on it?

Here is something you can do to test for stability.

Run Furmark, GPU-Z, and HWinfo64 to put a nice load on you core and measure GPU sensor info.

Here is my Sapphire HD7950 950mhz edition set at 1.162v with TRIXX(which I think is off +.04v) running Furmark for 40 minutes.
Ambient temperature of 29.5C.

ExtendedhottestingHD795040minutesfurmark.jpg



I used Trixx to set voltage but I think the Trixx software may be off a little on the voltage settings as GPU-Z and HWinfo64 pick up sensor reading voltage of around 1.06v under load and 1.128v without any load. Looks like a voltage drop of around .06v under load and Trixx is off by around +.04v.

Also does your CPU share the same loop?

I'm running i5-3570k at 4.9ghz in the same loop as my HD7950.
 
Last edited:
Damn you have a low voltage on the core.

highest i got on 1200mv was 1285 before it started locking up so looks like you have an insane card!!

and yup my CPU is on the same loop. Its not OCed yet, i wanted to get the GPU done first
 
Krayzie135 said:
Damn you have a low voltage on the core.

highest i got on 1200mv was 1285 before it started locking up so looks like you have an insane card!!

and yup my CPU is on the same loop. Its not OCed yet, i wanted to get the GPU done first
Click to expand...

Thanks. :cool:

I have the Sapphire 950mhz Edition HD7950 which is built on a reference HD7970 PCB. It's one of the newest HD7950s around, totally compatible with reference HD7970 water blocks. Amazon is selling them for $389.99 right now.

High over clocked CPU will increase overall temps in your loop as well.
 
hit 1300mhz on the core today, although i did have to give it 1245mv to keep it stable.

whats the max core voltage these things can handle for extended periods?
 
I have 2 more Powercolor reference cards hitting 1200/1600 with 1.25 volts. Liking these cards. Have not set all of them up in the same PC yet. Doubt I will run them at that high of a voltage together, but all individually have hit the 1200/1600 mark.
 
Krayzie135 said:
hit 1300mhz on the core today, although i did have to give it 1245mv to keep it stable.

whats the max core voltage these things can handle for extended periods?
Click to expand...

Not sure.

My voltage setting on my Sapphire 950mhz edition go all the way up to 1.38v but the fastest I have benchmarked my card is at 1350mhz at around 1.3v. AFter 1.3v my card gets very unstable and I'm not sure if it is because I have my CPU at 4.9GHZ along with a a water pump and huge fans and fan controller pinching my PSU.

I really think I may be getting into some type of PSU issues with my RIG when I'm pushing so many cross loads on it.

May have to RMA my PSU because running HWinfo64 my +3.3v measurements are nearing 3v under load. Seems like to much voltage loss.


Not sure how much voltage your card can handle. Could you do me a favor and run HWinfo64 and monitor your temps during furmark load? I'm curious as to how hot your VRM and GPU is getting under load.


Most VRM's are rated for around 120C but I don't think it is a good idea to run them that hot and I believe the heat heats up the PCB and effects other over all conductiveness of PCB for stability. Cooler is always better IMO.
 
What can I use to OC the XFX 7950 that was just the 'new hotness' on sale the last few days? It's stock at 800/1250.
 
SonDa5 said:
Not sure how much voltage your card can handle. Could you do me a favor and run HWinfo64 and monitor your temps during furmark load? I'm curious as to how hot your VRM and GPU is getting under load.


Most VRM's are rated for around 120C but I don't think it is a good idea to run them that hot and I believe the heat heats up the PCB and effects other over all conductiveness of PCB for stability. Cooler is always better IMO.
Click to expand...

yup ill test that right now.

what PSU are you running?

EDIT** 1300 core @ 1243mv, 1600 ram

1300mhz.jpg
 
Last edited:
Krayzie135 said:
yup ill test that right now.

what PSU are you running?
Click to expand...

Thanks.

Running Seasonic 760X which looks great on my 12v readings but my 3.5v readings are dipping as low as 3v. 500mv seems like way too much of a dip.

Ambient temperature 23.6C

1300mhztemps.jpg




Your block is doing a great job at cooling your VRMs. I did same test with your setting for the first 5 minutes then I tapered voltage down to 1.181 for another 5 minutes. You can see how the temps came down just a little. Looking at your 12v readings it looks like it is dipping and that might be why your vrms on your card are putting out alot less current. Extra current on my card VRMS is probably causing them to get extra hot as well.

Very impressed with your water block performance. Makes me double think about going with a gpu only Swiftech MCW82-7900. My card temps might be higher because my VRMS are putting out more amperage and my 12volt is higher. Not completely sure.
 
Last edited:
Any of you guys able to actually pass Unigine Heaven loops at max details for 60 minutes or so? I don't think anyone even uses Furmark anymore. Would like to see if these overclocks are stable in a current stress test.
 
Blkout said:
Any of you guys able to actually pass Unigine Heaven loops at max details for 60 minutes or so? I don't think anyone even uses Furmark anymore. Would like to see if these overclocks are stable in a current stress test.
Click to expand...


I have gamed UT3, BF3 and Crysis 2 for hours at these settings without any problems. I use Furmark to test because it is easy to run and does create a nice load and heat.

I think you had a decent HD7950 but not that great of a card. Over clocking with any card is kind of luck of the draw on over clocks. Also our cards are water cooled.
 
SonDa5 said:
I have gamed UT3, BF3 and Crysis 2 for hours at these settings without any problems. I use Furmark to test because it is easy to run and does create a nice load and heat.
Click to expand...

Heaven demo is free and provides a nice stability tool. Why not try it? I've seen many high clock speeds in the games you have listed but won't pass Heaven for 30-60 minutes. That may or may not matter to you but I think Heaven is the standard for which most people are stability testing right now, kind of hard to compare stable speeds to others when not using the same stability tools. Give it a try, its not like it costs you anything.

http://unigine.com/products/heaven/download/
 
Blkout said:
Heaven demo is free and provides a nice stability tool. Why not try it? I've seen many high clock speeds in the games you have listed but won't pass Heaven for 30-60 minutes. That may or may not matter to you but I think Heaven is the standard for which most people are stability testing right now, kind of hard to compare stable speeds to others when not using the same stability tools. Give it a try, its not like it costs you anything.

http://unigine.com/products/heaven/download/
Click to expand...

Over the last few weeks I've ran heaven at extreme settings and passed... Just didn't try it for an hour loop. I've ran it plenty of times for fun and my card doesn't heat up as much as Furmark. I don't see the point in running it for 1 hour. I'd rather play games for 1 hour.


Earlier this week I did 20 passes of LinX while running Furmark for a little over 1 hour. Only instability I saw was on my Seasoinc X760 that was showing over 400mv voltages loss on +3.5v line.

I was very impressed with my system performance over all. HD7950 matches up very nicely with Ivy Bridge. To me 20 passes of Linx on an over clocked CPU while running an over clocked GPU on Furmark is more impressive than 1 hour of Heaven benchmark.

29c ambient temperature.

STABLERIG35770kHD7950




Blkout why didn't you post in this thread back when you had your card? You make alot of negative noise about this card now but I don't see any screen shots of your card doing stress tests or any interest in this thread back when you had a HD7950.
 
Last edited:
Only had it a week, wasn't that impressed with the heat and noise it generated in order to overclock the card beyond 1100MHz. I already had one 670 in another machine that was performing just as well if not better in some games and ran 15-18c cooler and the fans were inaudible. The Sapphire cooler is not that impressive though. GIgabyte and Asus have far more impressive coolers with heatpipes that touch the core. The Sapphire had a block of copper and it was damn rough to touch, it just felt cheap and that's probably why it doesn't cool as well as a Gigabyte or Asus.

As for me posting screens, I don't have time for all that when I'm at home. I have a family and a 1 1/2 year old son that keeps me pretty busy. I don't get to see much computer time until very late in the evening when he's in bed and by then I'm too tired to run benches and post screens. I don't care if anyone believes me or not, I post my results for others to see, they can choose to believe or not. I've provided too much factual information on this board to not be credible.

I'm quite happy with my 670's in both machines, they perform every bit as well as a overclocked 7950 or 7970 and don't generate nearly as much heat doing it and since I gave up water cooling last year, I like a quiet air cooled heatink which the Sapphire 7950 OC was not under heavy load.
 
I have a family as well. You spend time sharing your opinions about how the GTX670 is better than HD7950 based on your 1 week experience with HD7950 but i have not had the same experience. I am having good experience with this card.

I hope other people have good experience with this card.
 
SonDa5 said:
I have a family as well. You spend time sharing your opinions about how the GTX670 is better than HD7950 based on your 1 week experience with HD7950 but i have not had the same experience. I am having good experience with this card.

I hope other people have good experience with this card.
Click to expand...

I realize its not really fair to you since you haven't owned both cards like I have. Its gives me a unique perspective to compare them.
 
Blkout said:
Heaven demo is free and provides a nice stability tool. Why not try it? I've seen many high clock speeds in the games you have listed but won't pass Heaven for 30-60 minutes. That may or may not matter to you but I think Heaven is the standard for which most people are stability testing right now, kind of hard to compare stable speeds to others when not using the same stability tools. Give it a try, its not like it costs you anything.

http://unigine.com/products/heaven/download/
Click to expand...

i have way too much time on my hands so i guess ill run the test for you sometime tonight;)

what settings were you looking for?
 
Last edited:
Krayzie135 said:
i have way too much time on my hands so i guess ill run the test for you sometime tonight;)

what settings were you looking for?
Click to expand...

No benches, just run Heaven at DX11, highest resolution your monitor supports, 8xAA, 16xAF, Extreme Tesselation. If it passes 60 minutes of that, its probably stable in anything. Its just difficult to compare similar overclocks when most people are using Heaven these days, and honestly, everyone should be, so when I see other clock speeds using other stability tests, its just hard to see if it would be the same using the Heaven standard.
 
I've got time to do the 1 hour Heaven run.


Just ran the benchmark mode at 1600/1300 1.22V under load.

xtessscoreheavenhd7950.jpg


Will keep the same settings and run it for 1 hour.

Will post screen shot of HWinfo64 so you can see how the card reacts to the load for an hour.
 
Over 2 hours of running Heaven. Stable.
Ambient Temp 25C.

swtcichtemps.gif
 
Last edited:
SonDa5 said:
I've got time to do the 1 hour Heaven run.


Just ran the benchmark mode at 1600/1300 1.22V under load.

Will keep the same settings and run it for 1 hour.

Will post screen shot of HWinfo64 so you can see how the card reacts to the load for an hour.
Click to expand...

That's great, at least you're using a stability tool that everyone else is using now. Its an apples to apples comparison. However the benchmark in Heaven is terribly unreliable, so I wouldn't even bother with it. You can run it 5 times and get wildly different results. I don't trust the benchmark in Heaven at all. However, just running loops at max settings is a fantastic stability tool which everyone should be using.

Sounds like you got a real winner with that card. Most people wouldn't have gotten that lucky, at least not at the voltage you're running.
 
Blkout said:
Sounds like you got a real winner with that card. Most people wouldn't have gotten that lucky, at least not at the voltage you're running.
Click to expand...

In this thread you stated that the highest you went on voltage was around 1.1v and then your card got too hot with quiet fan settings. If you would have water cooled it I think you could have gotten much better results.

The Sapphire HD7950 950mhz edition is built on a beefed up reference HD7970 PCB and is designed to over clock higher than normal HD7950s.

From Sapphire WEB site:

The new design is HD 7970 class that enables more head room for overclocking....On the OC edition, the second BIOS is pre-programmed to allow a higher maximum overclock and more aggressive fan profile..
Click to expand...

Did you try the 2nd BIOS switch position? Your comfortable over clock that you got at 1140/5800 OC at 1.093v really isn't bad and if you didn't try the 2nd BIOS switch that may be why your card wasn't as cool or stable at those settings. I think your card with a water block could have gotten much more faster with the 2nd BIOS swtich enabled.


I ran the Heaven loops double the time you requested just so that you can see that the card is stable and I posted the HWinfo64 sensor info so everyone can check out the sensor readings under load. My card is water cooled with a Swiftech MCW82-7900 gpu only block using the stock PCB heat sink. My Raven RV02 case has good air flow which I think helps keep the card cool.
 
Last edited:
Blkout said:
That's why everyone is using Heaven, its the best tool available right now for checking overclocks.
Click to expand...



I think running Furmark and running 20 passes of LinX at the same time is better.
 
SonDa5 said:
In this thread you stated that the highest you went on voltage was around 1.1v and then your card got too hot with quiet fan settings. If you would have water cooled it I think you could have gotten much better results.

The Sapphire HD7950 950mhz edition is built on a beefed up reference HD7970 PCB and is designed to over clock higher than normal HD7950s.

From Sapphire WEB site:



Did you try the 2nd BIOS switch position? Your comfortable over clock that you got at 1140/5800 OC at 1.093v really isn't bad and if you didn't try the 2nd BIOS switch that may be why your card wasn't as cool or stable at those settings. I think your card with a water block could have gotten much more faster with the 2nd BIOS swtich enabled.


I ran the Heaven loops double the time you requested just so that you can see that the card is stable and I posted the HWinfo64 sensor info so everyone can check out the sensor readings under load. My card is water cooled with a Swiftech MCW82-7900 gpu only block using the stock PCB heat sink. My Raven RV02 case has good air flow which I think helps keep the card cool.
Click to expand...

I believe my card obviously would have overclocked better under water, not many cards wouldn't but mine obviously required more volts for stability under Heaven than yours did too so like I said, you may have gotten lucky and your results certainly aren't the norm. After 1100MHz, it was starting to require too much voltage to keep stability. I managed 1140MHz before the voltage were getting out of hand for air cooling. Temps were simply too much for the Sapphire cooler to take at any reasonable fan speed. You didn't mind though as you ran yours at 100% fan speed, I don't consider that accpetable for day to day usage.

I've already stated you must have a great card, did you miss that part?

I tried both BIOS positions and both were exactly the same. I think you'll find most reviews will tell you the only difference if any betwen the switch positions is that one has a more aggressive fan profile setting but that didn't matter when I was using a custom fan profile that was more aggressive than the factory anyway.
 
Blkout said:
Not really as Krazie found out his card wasn't stable during Heaven but was during Furmark.
Click to expand...


He didn't run LinX for 20 passes while running Furmark. This takes at least an hour or more to do depending on how fast the CPU is over clocked.


He ran Furmark for 5 minutes.
 
i think something might be wrong with my card. With all the settings on full i get artifacts even at stock clocks...I have to drop the tessellation down to normal for it to run properly

driver/bios issues or is the card toast?
 
Krayzie135 said:
i think something might be wrong with my card. With all the settings on full i get artifacts even at stock clocks...I have to drop the tessellation down to normal for it to run properly

driver/bios issues or is the card toast?
Click to expand...

That doesn't sound good, might try checking your waterblock to make sure you're not shorting anything. Make sure you didn't overtighten it.
 
SonDa5 said:
He didn't run LinX for 20 passes while running Furmark. This takes at least an hour or more to do depending on how fast the CPU is over clocked.


He ran Furmark for 5 minutes.
Click to expand...

LinX is a CPU stability tool though, it has no bearing on a video card. By running both, all you did is slow down your video card since it was bottlenecked by your CPU.
 
Krayzie135 said:
i think something might be wrong with my card. With all the settings on full i get artifacts even at stock clocks...I have to drop the tessellation down to normal for it to run properly

driver/bios issues or is the card toast?
Click to expand...

On those HWinfo64 screen shots you posted earlier your 12v sensor readings were low. PSU could be problem.
 
Blkout said:
LinX is a CPU stability tool though, it has no bearing on a video card. By running both, all you did is slow down your video card since it was bottlenecked by your CPU.
Click to expand...



PLUS stressed total system.
 
SonDa5 said:
PLUS stressed total system.
Click to expand...

I understand that, but when trying to find the limits of a video card, testing the whole system isn't the best way to go about it. Should be testing one component at a time for stability then if you feel like you need to check the whole system, you do it last.

The problem with doing it your way is that the video card really isn't being pushed because the CPU can't feed it data fast enough because its being choked by LinX.
 
Blkout said:
I understand that, but when trying to find the limits of a video card, testing the whole system isn't the best way to go about it. Should be testing one component at a time.
Click to expand...

I do both. Running both time is neat to see.

For total system over clocking I have found that a slightly unstable CPU will not allow my HD7950 to go as fast. So having stable CPU over clock and stable GPU over clock are both important for finding total maximum performance for gaming or benchmarks.

Also when I run both I use HDinfo64 and monitor voltage sensors to find any weakness in my PSU or cooling.
 
You must log in or register to reply here.
Back
Top