Problems with GA-X79S-UP5 board - please help!

[sin0822]

It's not just happening on C2, my C0 exhibits the same issue.

From what I'm hearing, they are saying it's a CPU-related issue not a mainboard one. Did you swap out for another mainboard brand already? If yes - same issues?

//edit: couldn't wait for the reply, tested P9X79 with the same C0 chip and so far no throttling

//edit2: went back to the X79S-UP5 to test behavior with gear ratio at 1.25X. Now I don't have throttling at 4.25GHz anymore (34x multiplier), but around 4.75G (38x) it would throttle back to 37x once every minute or so (under high load). It seems like the throttling occurs more often and in larger steps the higher the multiplier you're using (and not related to temperature, voltage or frequency). I'll go back to the P9X79 to double check if no throttling takes place.

//edit3: back to P9X79 running 4.5G (45x) OCCT AVX-enabled and no throttling; on the same bench with the X79S-UP5 at 4.3G (43x) it'll drop back to 42x and 41x once every 15-20sec. As said, at 4.75G (38x) it only throttles back one step (37x) every 60sec or so. As far as I can tell, the engineers are working on it (or at least aware), but so far didn't see a bios that fixes the issue. I'll keep you updated.

Try BIOS f4r for the UP5 it should fix it for your C0 as it fixes it for my C0. BIOS engineers told me that the BIOS would fix for only 1 type of CPUs either ES or retail, when i tested it it fixed for C0 ES, so not retail.

[dejanh]

Try BIOS f4r for the UP5 it should fix it for your C0 as it fixes it for my C0. BIOS engineers told me that the BIOS would fix for only 1 type of CPUs either ES or retail, when i tested it it fixed for C0 ES, so not retail.

That's very useful to retail purchasers of the board. Also, you were real quiet for a while, but when I posted that massman picked this up all of a sudden you care again eh?

Don't take it personal, I'm just fed up with everyone dancing around this like it does not exist. It exists and it affects all of us who bought this board. Not a single piss-ant review out there admits that there is a problem, and all of my efforts on top of it all to actually get them to edit their reviews are fruitless because everyone is too afraid to say the truth about the quality. No wonder that every review out today for every product is a "9/10 amazing product".

[massman]

Heh ... "reviews", right /

I have tested other mainboards, namely RIVE. Blameless over on OCN also tested RIVE and P9X79 and none of those boards have issues running the same chip. I have tried multiple chips, multiple GA-X79S-UP5 boards, and multiple DRAM kits with different speeds and nothing makes a difference. This is definitely a problem with this mainboard, moreover I am convinced that it is firmware related. I also have direct confirmations of issues from all other users of this board. They all have the same exact problem.

Gigabyte really dropped the ball on this one. The thing is though that in my communication with them it's now been over 1.5 months with this issue and they are still refusing to admit to me that there is a widespread problem (which there clearly is since it affects all users). They are constantly trying to blame in on my "chip quality" and no matter how much proof I provide to them they just keep going back to the same thing over and over again. I even recorded a video for them for the love of god!

"It's only in some cases" they say, "it's only over 4.4GHz" they say, male bovine excrement I say.

Do you have any updates from them?

No updates, but they are most certainly aware of the problem. As far as I know, there is a long internal email trail to figure out the issue.

Normally, this kind of throttling would either be triggered by the VRM or CPU internal OCP but both having VRM under LN2 (temp) and disabling cores doesn't resolve the problem. But as I mentioned in my previous post: with higher voltage and changing to 1.25x BCLK, I don't get the throttling at 4.3G, even though the power consumption should be higher. So, I reckon it's something else ... but I don't know what. I'm really interested in the answer, though .

It's not in "some cases" and not "only over 4.4G".

Try BIOS f4r for the UP5 it should fix it for your C0 as it fixes it for my C0. BIOS engineers told me that the BIOS would fix for only 1 type of CPUs either ES or retail, when i tested it it fixed for C0 ES, so not retail.

Well, just tested that BIOS and it doesn't so you shouldn't tell them it's fixed . F4r, F4s and F4t all exhibit the same issue.

//edit: well, seems like a solution is in-bound ... well, an interesting solution that is. More info today, I think.

[dejanh]

Heh ... "reviews", right /



No updates, but they are most certainly aware of the problem. As far as I know, there is a long internal email trail to figure out the issue.

Normally, this kind of throttling would either be triggered by the VRM or CPU internal OCP but both having VRM under LN2 (temp) and disabling cores doesn't resolve the problem. But as I mentioned in my previous post: with higher voltage and changing to 1.25x BCLK, I don't get the throttling at 4.3G, even though the power consumption should be higher. So, I reckon it's something else ... but I don't know what. I'm really interested in the answer, though .

It's not in "some cases" and not "only over 4.4G".



Well, just tested that BIOS and it doesn't so you shouldn't tell them it's fixed . F4r, F4s and F4t all exhibit the same issue.

//edit: well, seems like a solution is in-bound ... well, an interesting solution that is. More info today, I think.

Now you got me excited Can't wait to see what was devised and especially what is the cause Looking forward to that update!

By the way, in my tests I did manage to stop the throttling by disabling cores, or rather by specifying in Prime95 Small FFT test or LinX to use less cores. However, the higher the voltage or the overclock, or both, the less cores and threads you have to use to avoid seeing throttling. So maybe at 4.4GHz I'm able to stop throttling by using only 3C/6T in Prime95 Small FFT but then at 4.6GHz I have to drop that to 2C/4T only. With that said, there could still be throttling occurring but it could be very quick so that I do not necessarily see the transitions any more. Just thought I'd share that observation. I'm pretty sure that at some point in time I documented it in one of the other threads...

I also did not observe throttling when AVX is not used, but that's not to say that it would not happen with a high enough overclock/voltage combination even without AVX. Not using AVX just reduces the overall current/power demands so it makes sense why throttling stops (at least to an extent).

[sin0822]

Now you got me excited Can't wait to see what was devised and especially what is the cause Looking forward to that update!

By the way, in my tests I did manage to stop the throttling by disabling cores, or rather by specifying in Prime95 Small FFT test or LinX to use less cores. However, the higher the voltage or the overclock, or both, the less cores and threads you have to use to avoid seeing throttling. So maybe at 4.4GHz I'm able to stop throttling by using only 3C/6T in Prime95 Small FFT but then at 4.6GHz I have to drop that to 2C/4T only. With that said, there could still be throttling occurring but it could be very quick so that I do not necessarily see the transitions any more. Just thought I'd share that observation. I'm pretty sure that at some point in time I documented it in one of the other threads...

I also did not observe throttling when AVX is not used, but that's not to say that it would not happen with a high enough overclock/voltage combination even without AVX. Not using AVX just reduces the overall current/power demands so it makes sense why throttling stops (at least to an extent).

well that is different b/c my C0 would exhibit throttling without AVX......


With AVX my C0 CPus just freeze at higher speeds and crash, they can't handle being stable at any voltage at 4.5ghz+.

I also did try without AVX to compare with F4R and previous, and it worked(IBT non AVX). Then i tried with F4R with LinX and the system kept freezing after a few loops, other BIOSes would show throttling and no freezing, thus F4R was stopping the throttling but the CPU couldn't take it, so I measured the input power on the 12v rails and it did confirm that the power didn't drop like it did with throttling with F4R.

i will make a video, but the issue is that people will think that i am showing the issue be fixed.

[dejanh]

well that is different b/c my C0 would exhibit throttling without AVX......


With AVX my C0 CPus just freeze at higher speeds and crash, they can't handle being stable at any voltage at 4.5ghz+.

I also did try without AVX to compare with F4R and previous, and it worked(IBT non AVX). Then i tried with F4R with LinX and the system kept freezing after a few loops, other BIOSes would show throttling and no freezing, thus F4R was stopping the throttling but the CPU couldn't take it, so I measured the input power on the 12v rails and it did confirm that the power didn't drop like it did with throttling with F4R.

i will make a video, but the issue is that people will think that i am showing the issue be fixed.

Thanks for that feedback. It's certainly interesting and it definitely illustrates that this is a rather peculiar issue that we are dealing with. I always keep my Kill-A-Watt meter plugged in between my wall and the computer and (without measuring the 12V rail) I do definitely see a power drop when not using AVX which is consistent with the software-based power reporting, and with what I would intuitively expect to see. The behavior that I am seeing is at least consistent with all of the retail chips that people have reported to date. At the end of the day, it's these ones that need to be fixed primarily as that is 99% user base for this product.

[ieatrawbacos]

Normally, this kind of throttling would either be triggered by the VRM or CPU internal OCP but both having VRM under LN2 (temp) and disabling cores doesn't resolve the problem. But as I mentioned in my previous post: with higher voltage and changing to 1.25x BCLK, I don't get the throttling at 4.3G, even though the power consumption should be higher. So, I reckon it's something else ... but I don't know what. I'm really interested in the answer, though .

The throttling seems tied to the turbo multipliers, which apply current and wattage limits that do not exist on non-turbo ratios, and are more generous with lower turbo ratios, on LGA-2011 parts. The 1.25x strap allows you to use a much lower multiplier that is much less likely to be throttled.

I was also able to get my X79S-UP5 to be throttle free using 1.25x, but 1.25x is barely usable with my chip on air (horrible cold boot issues, so I need multiplier restarts bumping multiplier a few notches at a time, and even then it's not stable), so it's not at all a practical solution.

//edit: well, seems like a solution is in-bound ... well, an interesting solution that is. More info today, I think.

Any more info on this?

On a related tangent, I remember on of the fixes ASUS applied to turbo ratios with their LGA-1366 P6T line. They initially left the TDP limit that throttled the turbo multipler off at stock values, which caused serious issues with i7 920s (as odd multipliers clocked better, and 21x was the turbo ratio). So what it seems they did was just cut the reported power consumption in half. It was a half-assed solution, but it worked.