Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
1800? Man, we are gonna see some dead cards soon Does anyone really think 1Ghz OC on an HD5870 is possible??
Lots are hitting 1000 core.
Stock clock here.
1000/1300 bench here. Weird CPUID doesn't see my GPU voltage increase.
Did you flash your bios to break the CCC limit?
Yup.
Got a link to the bios and flash tool used for it?
the performance drops with these anyways. Sometimes more isnt always better.
I can't go further for some reason.
Probably because your card is getting so hot it is melting.
1000/1300 bench here. Weird CPUID doesn't see my GPU voltage increase.
I don't think the 40nm chips like to be above 60 degrees.
I keep my 40nm chip idling in the low 20's and under stress no higher than low 40's.
In short the card is getting to hot. 40nm die is just way to small. Can't handle the heat. Also since it is so small that is why it is so sensitive to voltage. Need better cooling for better over clocking.
Raising the core will also give a performance degrade. The core doesn't operate "fine fine fine BAM crash;" as you get closer to the stability limits, the performance will degrade and then it will crash.The 5870 can handle up to 90C before damage becomes a concern. Thermal protection only kicks in close to 100C. In short, what you think is just plain wrong. If these chips couldn't handle the heat, AMD wouldn't have released cards that usually top off at around 80C at stock settings.
To vjcsmoke: Raising the core speed too high will only result in hard locks and crashes, not a performance degrade. That only occurs with the memory. Testing with my card, I could take the core all the way up to 1045 with performance increases, and at 1050, it would hard-lock. Now with memory, anything past 1300 did result in lower performance, but only single FPS or two. You'd have to be looking hard to notice it.
No, heat is exactly the problem with these cards. You can use much less voltage at moderately lower temps. If you can keep the 5870 below 70C load, 1GHz with less than 1.3V is almost guaranteed.The real issue isn't heat on the cards, they just need more voltage to hit the higher frequencies with stability.
However, too much voltage and you both decrease the life of your card, and you may need to hard mode the card in some way.
Your 3dmarks are 17.7k at that speed which is good, but... I get 19k 3dmarks at 975/1300. Hmmm? Does my extra 100mhz of CPU speed make that big of a difference or are we seeing subtle performance degradation at the higher clock?
And obviously I get 21K 3dmarks with my 9600GSO enabled for hardware PhysX but that's almost unfair advantage cos of 3dmarks PhysX bias so we won't go into that.
Raising the core will also give a performance degrade. The core doesn't operate "fine fine fine BAM crash;" as you get closer to the stability limits, the performance will degrade and then it will crash.
No, heat is exactly the problem with these cards. You can use much less voltage at moderately lower temps. If you can keep the 5870 below 70C load, 1GHz with less than 1.3V is almost guaranteed.
Raising my Core alone never gave me a performance decrease. Like I said, 1045 at 1.27v is perfectly stable while 1050 would hard-lock after a couple seconds, or the card would crash to desktop as soon as I tried to start a game. The games I'm using for testing are 1).Crysis 2).Far Cry 2 3).Red Faction:Guerrilla and 4).Need for Speed:Shift. I'm guessing my card could be different than the majority, since I'm not limited by temperature on my overclocks. Even running the fan at 75%, which would keeps my temps around 65C, would not make 1050 stable. If temperature was a limit, than that should have allowed me to get there. My card seems totally dependent on voltage. In fact, as I type this, my 5870 is running 1050/1300 on 1.287 and I'm also using the default fan profile. Temps have spiked at 88C, but it's still in running order. The only bad thing is that there are no games that are going to require this much shader power. Here's to hoping that Rage will scale well graphically and push our rigs. Something needs to bump of Crysis of its pedestal of "Best Looking Game".
As the week goes on, I'll start pushing my card to it's maximum, and if I see any of the symptoms you're describing, I'll let you know. But, I think that if they were going to show up, I would have seen something by now.
Show me a screen shot id like to compare. Your also running an i7 and I am on a Q9650. From the other screen shots I have seen I am dead on for 1000/1300.. yours seems high to me. I know one thing.. Temps play a big role in stability.. I got over 80C even on 975/1300 and she locked on me. Little fan adjustments fixed it though.
Does the Asus bios change the default voltages too?
Asus is supposed to come out with that card over 1000 on the core, do we know if thats is software or a hardware solution?
I dont want to have to mess with the MSI software still if I can get a bios with slightly higher voltages too.
I found its easy to pass Vantage with even an unstable overclock. Try passing a few benches of Furmark in a row or some heavy hours of gaming. That is when I found my 1ghz was not stable even at 1.25V. What was causing the instability was heat. If I jacked the fan up to 80% I could remain stable. I was getting crashes even at 975/1300 when the temps got up over 80C. Long as I kept them under that it was all fine.
Show me a screen shot id like to compare. Your also running an i7 and I am on a Q9650. From the other screen shots I have seen I am dead on for 1000/1300.. yours seems high to me. I know one thing.. Temps play a big role in stability.. I got over 80C even on 975/1300 and she locked on me. Little fan adjustments fixed it though.
And the performance degradation is at 1050, yes? It's not the same in every application, and sometimes it can happen over 5MHz. As the GPU becomes unstable, performance will start to drop until it's pushed passed the point of functionality and crashes. And it's temperature or voltage, it's a combination of both. You can't just get the card cool and clock it to the moon just like you can't set it to 1.35V and get it high either. Offhand, it sounds like you got a good piece of silicon if it can run those speeds at those temps. But notice how you need 1.287V to keep it stable? At cooler temps, you wouldn't need nearly that much voltage. For example:Raising my Core alone never gave me a performance decrease. Like I said, 1045 at 1.27v is perfectly stable while 1050 would hard-lock after a couple seconds, or the card would crash to desktop as soon as I tried to start a game. The games I'm using for testing are 1).Crysis 2).Far Cry 2 3).Red Faction:Guerrilla and 4).Need for Speed:Shift. I'm guessing my card could be different than the majority, since I'm not limited by temperature on my overclocks. Even running the fan at 75%, which would keeps my temps around 65C, would not make 1050 stable. If temperature was a limit, than that should have allowed me to get there. My card seems totally dependent on voltage. In fact, as I type this, my 5870 is running 1050/1300 on 1.287 and I'm also using the default fan profile. Temps have spiked at 88C, but it's still in running order. The only bad thing is that there are no games that are going to require this much shader power. Here's to hoping that Rage will scale well graphically and push our rigs. Something needs to bump of Crysis of its pedestal of "Best Looking Game".
As the week goes on, I'll start pushing my card to it's maximum, and if I see any of the symptoms you're describing, I'll let you know. But, I think that if they were going to show up, I would have seen something by now.
Exactly, I am not one bit surprised .1125mv was stock. Had to bump to 1160mv for stability @ 1Ghz. Water cooled of course.
http://www.hardforum.com/showthread.php?t=1457011
*snip*
And the performance degradation is at 1050, yes? It's not the same in every application, and sometimes it can happen over 5MHz. As the GPU becomes unstable, performance will start to drop until it's pushed passed the point of functionality and crashes. And it's temperature or voltage, it's a combination of both. You can't just get the card cool and clock it to the moon just like you can't set it to 1.35V and get it high either. Offhand, it sounds like you got a good piece of silicon if it can run those speeds at those temps. But notice how you need 1.287V to keep it stable? At cooler temps, you wouldn't need nearly that much voltage. For example:
Exactly, I am not one bit surprised .
Here's my 19k screenshot with hardware PhysX disabled. 5870 @ 975/1300. i7 @ 4.1Ghz. Fans are on auto control.
Same speeds as above but with PhysX enabled.
I don't think the 40nm chips like to be above 60 degrees.
I keep my 40nm chip idling in the low 20's and under stress no higher than low 40's.
In short the card is getting to hot. 40nm die is just way to small. Can't handle the heat. Also since it is so small that is why it is so sensitive to voltage. Need better cooling for better over clocking.
The 5870 can handle up to 90C before damage becomes a concern. Thermal protection only kicks in close to 100C. In short, what you think is just plain wrong. If these chips couldn't handle the heat, AMD wouldn't have released cards that usually top off at around 80C at stock settings.
Nice to see overclocked HD5870 at ~1000/1300 core/mem pulling a pretty consistent GPU score of 18000-18500 in Vantage.Ahh your CPU score is much higher then mine. Explains some of the difference. My GPU score and FPS scores are almost identical though with the same clock on the video card.
Wow you really just have no idea what you are talking about do you...
+1