Poll: Ryzen 7950X3D, 7900X3D, 7800X3D

[LtMatt]

The only time I've had a issue with a game not going onto the 3d chiplet was the Once Human alpha as game bar didn't detect it automatically, ticked the remember this is a game and was all ok.

Yes. If you don't keep Game Bar up to date this (remember this is a game) may be required for new titles. I've also seen it required when a new game launches and game bar doesn't get updated for a few days.

Once game bar is updated and the game is profiled, game bar will pick the best CCD for the game.

That said, it's not as if the game will run bad if it uses the frequency CCD. Most people won't even notice unless you study an overlay and check which cores are active with work while gaming.

 

[Audioboxer]

Replacement CPU is goosed as well

Everything on default above.

PBO turned on, +10 positive curve on all cores.

Hopefully OCUK will just give me a refund now, I'm not playing the lottery any more with 7950x3Ds.

 

[the wopster]

Replacement CPU is goosed as well

Everything on default above.

PBO turned on, +10 positive curve on all cores.

Hopefully OCUK will just give me a refund now, I'm not playing the lottery any more with 7950x3Ds.

Is this not suggesting that OCCT is the problem, are you failing other stress tests and benchmarks?

 

[Audioboxer]

Is this not suggesting that OCCT is the problem, are you failing other stress tests and benchmarks?

Any AVX2 with Prime95. You can do the same tests with Corecycler rather than OCCT.

If I can get a refund accepted rather than another replacement I'll probably just buy a 7800x3D.

 

[LtMatt]

Any AVX2 with Prime95. You can do the same tests with Corecycler rather than OCCT.

If I can get a refund accepted rather than another replacement I'll probably just buy a 7800x3D.

Can you show the same errors on this new CPU with Prime95 running core cycler testing each core for 1 minute?

For me OCCT threw errors at stock on multiple 7950X3Ds too, yet Core cycler and Prime95 would run for days without issue. The system was stable too (unlike yours with the previous CPU which you said was unstable even at idle on desktop). I never got to the bottom of it, for me OCCT was showing false errors that I couldn't reproduce with Prime 95 AVX2. I gave up testing with it in the end as it didn't seem reliable. OCCT was okay in other tests, it was just that short cycle test where each core gets tested for like a few seconds before it moves onto the next. Something about that would just throw seemingly false errors with everything at stock and memory at 5200Mhz.

 

[Audioboxer]

Can you show the same errors on this new CPU with Prime95 running core cycler testing each core for 1 minute?

For me OCCT threw errors at stock on multiple 7950X3Ds too, yet Core cycler and Prime95 would run for days without issue. The system was stable too (unlike yours with the previous CPU which you said was unstable even at idle on desktop). I never got to the bottom of it, for me OCCT was showing false errors that I couldn't reproduce with Prime 95 AVX2. I gave up testing with it in the end as it didn't seem reliable. OCCT was okay in other tests, it was just that short cycle test where each core gets tested for like a few seconds before it moves onto the next. Something about that would just throw seemingly false errors with everything at stock and memory at 5200Mhz.

Interesting you reply with that because

All I changed was memory to 6000 and FCLK to 2200. I was mainly wanting to test FCLK booting at 2200 on this chip.

I think the VDDP has auto'd a bit higher, likely with the memory frequency bump at 1:1. VDDP possibly having an impact on AVX2 stability at stock? I'll go back to total stock and see if it drops down.

Corecycler AVX2 was crashing like mad on my other CPU, I'll test that on this one.

Ignore this, the default profile I saved in the BIOS that I thought only had AURA turned off/some onboard devices turned off for quickness also had PBO +10 enabled lol. OCCT still crashes like mad on default/PBO off.

 

[nlel1975]

FWIW mine does it too in OCCT doesn’t bother me as my machine appears to be stable otherwise.

 

[LtMatt]

FWIW mine does it too in OCCT doesn’t bother me as my machine appears to be stable otherwise.

Yeah. I am almost convinced that for most it's false errors as I could not reproduce instability in any other test. I feel bad because i actually returned a couple of CPUs that I now believe were not faulty at all, including one great sample.

 

[Audioboxer]

FWIW mine does it too in OCCT doesn’t bother me as my machine appears to be stable otherwise.

Does it do it in Corecycler? Mine does as well.

Can you show the same errors on this new CPU with Prime95 running core cycler testing each core for 1 minute?

For me OCCT threw errors at stock on multiple 7950X3Ds too, yet Core cycler and Prime95 would run for days without issue. The system was stable too (unlike yours with the previous CPU which you said was unstable even at idle on desktop). I never got to the bottom of it, for me OCCT was showing false errors that I couldn't reproduce with Prime 95 AVX2. I gave up testing with it in the end as it didn't seem reliable. OCCT was okay in other tests, it was just that short cycle test where each core gets tested for like a few seconds before it moves onto the next. Something about that would just throw seemingly false errors with everything at stock and memory at 5200Mhz.

Corecycler brings the errors

Same as other CPU I sent back.

Ignore the original post above, I edited it out and explained what was going on (tldr; +10 positive curve was on).

Yeah. I am almost convinced that for most it's false errors as I could not reproduce instability in any other test. I feel bad because i actually returned a couple of CPUs that I now believe were not faulty at all, including one great sample.

Issue is other people's 7950x3Ds pass AVX2 no problem

 

[nlel1975]

Does it do it in Corecycler? Mine does as well.

I’ve only tested in OCCT in the way you mentioned, I think everyone with the 7950X3D has it. Those who say they don’t are lying… to themselves most likely.

 

[Audioboxer]

I’ve only tested in OCCT in the way you mentioned, I think everyone with the 7950X3D has it. Those who say they don’t are lying… to themselves most likely.

Sadly not, there is quite a bit of evidence on the Overclock.net forums of OCCT AVX2 being passed, some even with pretty decent negative curve offsets.

 

[nlel1975]

Sadly not, there is quite a bit of evidence on the Overclock.net forums of OCCT AVX2 being passed, some even with pretty decent negative curve offsets.

Well for me as my machine is otherwise stable, it’s not something I’m going to RMA over.

 

[Audioboxer]

Well for me as my machine is otherwise stable, it’s not something I’m going to RMA over.

Did you manage to get a negative offset on your best cores on CCD1? My main issue other than AVX2 was cores needing a positive offset when PBO was on.

Those were the same cores that also needed a positive offset to pass AVX2 as well.

 

[nlel1975]

Did you manage to get a negative offset on your best cores on CCD1? My main issue other than AVX2 was cores needing a positive offset when PBO was on.

Those were the same cores that also needed a positive offset to pass AVX2 as well.

To be honest I’ve not even tried to tweak it to make it pass.

 

[Audioboxer]

To be honest I’ve not even tried to tweak it to make it pass.

As in you don't use PBO or you simply haven't bothered to change your PBO values to try and make it pass? It's usually your best two cores on the frequency chip that trip up AVX2 in OCCT. Just put both of them on +10 and it'll likely work lol.

 

[nlel1975]

As in you don't use PBO or you simply haven't bothered to change your PBO values to try and make it pass? It's usually your best two cores on the frequency chip that trip up AVX2 in OCCT. Just put both of them on +10 and it'll likely work lol.

No I haven’t gone into the BIOS to make any changes, just running it at stock. Only things I’ve done to the BIOS is update it to latest version and set the ram to EXPO 1. I’ve also enabled Memory Context Restore.

Didn’t update the BIOS because of this, but because of the AMD RAM training issue which has been solved by enabling Memory Context Restore.

 

[henry hoov3r]

AMD RAM training issue which has been solved by enabling Memory Context Restore.

Does speed up the boot times? Im still on a pretty old bios.

 

[nlel1975]

Does speed up the boot times? Im still on a pretty old bios.

Yes, the BIOS update made no difference but enabling Memory Context Restore sped them up considerably. Now it boots just like my old AM4 DDR4 system.

 

[McCarthy]

I think if you can pass in all tests apart from one (OCCT in this case) then fair enough you can put it down to the test. Failing everything stock is unacceptable. I would disagree that you should be expected to add to certain cores for example to pass testing.

There have been many people with this issue unfortunately. Most exchange and have no problems with another sample.

Very unlucky.

 

[Audioboxer]

I think if you can pass in all tests apart from one (OCCT in this case) then fair enough you can put it down to the test. Failing everything stock is unacceptable. I would disagree that you should be expected to add to certain cores for example to pass testing.

There have been many people with this issue unfortunately. Most exchange and have no problems with another sample.

Very unlucky.

Yeah just an update to say I managed to get a CPU that passes AVX2 at stock, both in OCCT and Corecycler and this is with the BIOS defaulted. So they do exist. Makes it frustrating therefore that AMD is pushing out CPUs they must know have cores that won't make these tests at stock.

Like yourself I would say if something is crashing at stock and it's repeatable across multiple stability testing apps it's not really acceptable. Especially if you get some random reboots.