Ryzen Owners Zen Garden

[trparky]

Still problems after disabling Memory Context Restore?

No issues since disabling it, at least from what I can see. I've not rebooted in some time. I mean, I could reboot now and find out if it comes back up without having to play in UEFI again.

puts some actual load on the hardware.

But I thought that a memory test like Windows Memory Diagnostics hammers memory with several types of tests to determine if any of the memory cells are bad.

Performed a warm boot, came back up with no issues. I then powered the system down, shut off power at the switch at the back of the power supply and held the power button on the front of the case down for some time just to make sure that I drained any and all capacitors on the motherboard. I then flipped the switch at the back of the power supply and booted back up. Again, no issues.

It seems that even with Memory Context Restore disabled the boot times are nowhere near as bad as it was under the old BIOS version. Perhaps keeping Memory Context Restore disabled is a good thing and shouldn't even be enabled if it only causes problems and doesn't cause boot length issues.

 

[tabascosauz]

No issues since disabling it, at least from what I can see. I've not rebooted in some time. I mean, I could reboot now and find out if it comes back up without having to play in UEFI again.

But I thought that a memory test like Windows Memory Diagnostics hammers memory with several types of tests to determine if any of the memory cells are bad.

The point is to hit CPU/UMC/memory equally hard, and to do so for an extended period of time. Have you ever actually watched HWInfo during TM5 or Linpack?

Windows Memory Diagnostic is in the same boat as Memtest86 from a USB. Entire CPU is running basically idle at 30-35C. Borderline worthless except for defective hardware. Even less stressful/effective tests like HCI or P95 Large FFT at least run the UMC hard, and temps will visibly reflect that (ie. ~50C Tdie).

 

[trparky]

The point is to hit CPU/UMC/memory equally hard, and to do so for an extended period of time. Have you ever actually watched HWInfo during TM5 or Linpack?

Windows Memory Diagnostic is in the same boat as Memtest86 from a USB. Entire CPU is running basically idle at 30-35C. Borderline worthless except for defective hardware. Even less stressful/effective tests like HCI or P95 Large FFT at least run the UMC hard, and temps will visibly reflect that (ie. ~50C Tdie).

OK, I get it. I assumed that you meant just find out if the memory modules are bad.

 

[JrRacinFan]

The freq ceiling for the Vcache CCD should be around 5.1-5.2 across all the Raphael Vcache products. Any higher needs to come from the other CCD.

Make sure you have both the new BIOS/agesa and the new chipset drivers for the 2CCD optimizations

Rhetorical q...
Then why in the eff does it say up to 5.6Ghz on the shi.....

 

[3x0]

Then why in the eff does it say up to 5.6Ghz on the shi.....

Because programs utilizing the second non 3d vcache ccd will boost up to 5.6

 

[tabascosauz]

Rhetorical q...
Then why in the eff does it say up to 5.6Ghz on the shi.....

Remember when Ryzen 3000 launch AMD said "up to 4.4GHz"? Yeah, I also remember the 7900X3D and 7950X3D are very hands-on if you want 100% correct functionality and there's no way around that.

Where are you testing your ST boost clock? Cinebench? With latest BIOS and driver it'd be surprising if the scheduler still couldn't handle R23 of all things.

Also, at launch 7900X3D and 7950X3D seem to be falling slightly short of their spec Fmax even on the correct CCD, might be more AGESA fixes incoming

OK, I get it. I assumed that you meant just find out if the memory modules are bad.

Actually bad hardware is rare, and wouldn't have gone under the radar for this long. When say memtesting, it's for stability of whatever profile you are running

 

[JrRacinFan]

@tabascosauz take my posts with a grain of salt (or maybe even a pound lol) it's more less me ranting

 

[tabascosauz]

@tabascosauz take my posts with a grain of salt (or maybe even a pound lol) it's more less me ranting

Don't blame you at all, been there many a time with AMD

There's good info on page 5 of the 7950X3D review though worth a read, I'm sure you still want to be getting all the performance you paid for

 

[trparky]

Linpack Xtreme v1.1.5 by Regeneration (64-bit)

Current date/time: Fri Mar 03 18:56:41 2023

CPU frequency:    5.412 GHz
Number of CPUs: 16
Number of cores: 16
Number of threads: 16

Parameters are set to:

Number of tests: 1
Number of equations to solve (problem size) : 35000
Leading dimension of array                  : 35000
Number of trials to run                     : 10
Data alignment value (in Kbytes)            : 4

Maximum memory requested that can be used=1210769504, at the size=35000

=================== Timing linear equation system solver ===================

Size   LDA    Align. Time(s)    GFlops   Residual     Residual(norm) Check
35000  35000  4      68.645     416.4295 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.941     414.6422 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.832     415.2999 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.831     415.3025 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.846     415.2144 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.924     414.7457 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.929     414.7148 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.847     415.2078 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.947     414.6071 1.077145e-009 3.126789e-002   pass
35000  35000  4      68.768     415.6814 1.077145e-009 3.126789e-002   pass

Performance Summary (GFlops)

Size   LDA    Align.  Average  Maximal
35000  35000  4       415.1845 416.4295

Residual checks PASSED

End of tests

 

[freeagent]

That is on an 8 core part? I expected a bit more uplift. Still pretty good though.

 

[trparky]

That is on an 8 core part? I expected a bit more uplift. Still pretty good though.

Uh... excuse me? Please explain.

 

[freeagent]

Uh... excuse me? Please explain.

My 58X3D does 334GFlops and my 5900X does~520GFlops.

Not bad for an 8 core.

 

[Kapone33]

I have to say that I am totally enjoying my 7900X3D and have some thoughts on why this chip was not sampled to reviewers.

1. In some ways this is just as fast as the 7950X3D from comparing benchmarks
2. In Gaming in some cases it is faster than the 7950X3D (Could be due to temps)
3. It boosts to 5.64 GHZ which is not too far from 5.7
4.It runs cooler than the 5800X3D but in some Games is not faster due to 6 core Complex (But not slower)
5.Like the 5900X it is rock solid in RAM EXPO support
6. Mine was $799 Canadian vs the $1079 for the 7950X3D
7. It utilizes VRAM much better than the 5800X3D as I see as much as 13.5 GB VRAM usage playing TWWH3. It never went above 11 before that.
8. The chip is about 80% the size of the 5800X3D with more cores and much higher boost clocks.
9. It has a GPU attached too.
10. It does not go above 60 C.

I have it paired with a 7900XT. I would show you how sweet Gaming is with this chip but to get an idea of what I am talking about here are my Firestrike scores.

It is not all roses though.

1. It is $799 and the 5800X3D is $439 so that is $300
2. The boards with the highest flexibility are the most expensive and in some cases are double their AM4 variant.
3.The CPU does spike to 82 C when doing some tasks.
4. There is no noticeable difference in speed running M2 over Pcie 5.

Having said all of that you don't need an expensive AM5 board to enjoy this chip but the PCIe allocation is better than any X570 board.

 

[freeagent]

1. It is $799 and the 5800X3D is $439 so that is $300
2. The boards with the highest flexibility are the most expensive and in some cases are double their AM4 variant.
3.The CPU does spike to 82 C when doing some tasks.
4. There is no noticeable difference in speed running M2 over Pcie 5.

I paid the same price for my 5900X when it came out, seems fair.
That's usually how it is.
Then it doesn't run cooler than a 58X3D. Mine is super easy to cool, like the rest of my Vermeer parts.. as my 58X3D sits at 80c at 4250 running p95 small ffts.

 

[JrRacinFan]

. It boosts to 5.64 GHZ which is not too far from 5.7

Can you post an HWINFO screenshot please. I'm rather curious

 

[Kapone33]

I paid the same price for my 5900X when it came out, seems fair.
That's usually how it is.
Then it doesn't run cooler than a 58X3D. Mine is super easy to cool, like the rest of my Vermeer parts.. as my 58X3D sits at 80c at 4250 running p95 small ffts.

I am not saying that it is not easy to cool but remember that the 5800X3D does not have a GPU attached. The 82 spike is just that it does not sustain that temp for more than 5 seconds. I will run Cinebench tomorrow and see what kind of temps I see. There is a 2-4 degree difference in temps vs the CCDs as well in HWinfo.

Can you post an HWINFO screenshot please. I'm rather curious

 

[JrRacinFan]

@kapone32
And core effective clocks please? I think I may have to reseat my cpu's water block

 

[tabascosauz]

8. The chip is about 80% the size of the 5800X3D with more cores and much higher boost clocks.

1. It is $799 and the 5800X3D is $439 so that is $300

I mean, you're comparing the 5800X3D's current pricing; if you picked one up when it came out, it certainly would have been about $600cad. Probably similar price to what the 7800X3D will command next month.

$800 doesn't seem all that bad, enjoy the new hardware.

Not sure where the 80% is coming from though:

 

[Kapone33]

@kapone32
And core effective clocks please? I think I may have to reseat my cpu's water block

I mean, you're comparing the 5800X3D's current pricing; if you picked one up when it came out, it certainly would have been about $600cad. Probably similar price to what the 7800X3D will command next month.

$800 doesn't seem all that bad, enjoy the new hardware.

Not sure where the 80% is coming from though:

I am talking about the size of the physical chip. If you can get Wizz to post of a pick of a 5000 chip beside a 7000 chip and it will clearly show that the 7000 chip is indeed smaller than the 5000. To me it is a marvel. I am not going to take my CPU out to show it but when I was changing the system I did compare the CPUs and the 5000 chip is bigger.

I mean, you're comparing the 5800X3D's current pricing; if you picked one up when it came out, it certainly would have been about $600cad. Probably similar price to what the 7800X3D will command next month.

$800 doesn't seem all that bad, enjoy the new hardware.

Not sure where the 80% is coming from though:

I am talking about the size of the physical chip. If you can get Wizz to post of a pick of a 5000 chip beside a 7000 chip and it will clearly show that the 7000 chip is indeed smaller than the 5000. To me it is a marvel. I am not going to take my CPU out to show it but when I was changing the system I did compare the CPUs and the 5000 chip is bigger.

 

[tabascosauz]

I am talking about the size of the physical chip. If you can get Wizz to post of a pick of a 5000 chip beside a 7000 chip and it will clearly show that the 7000 chip is indeed smaller than the 5000. To me it is a marvel. I am not going to take my CPU out to show it but when I was changing the system I did compare the CPUs and the 5000 chip is bigger.

Are you referring to the smaller IHS size minus the "legs"? AM5 substrate is 40x40mm and AM4 is also 40x40mm, that's why all the die shot estimations and comparisons are possible.

 

[trparky]

My 58X3D does 334GFlops and my 5900X does~520GFlops.

Not bad for an 8 core.

OK, I get it now.

All I know was that it was a massive upgrade from my old 8700K.

 

[freeagent]

mean, you're comparing the 5800X3D's current pricing; if you picked one up when it came out, it certainly would have been about $600cad

Yeah I paid 600 for my 58X3D. And then 4 or 5 months later they lowered the price by 200.

Needless to say that I am less than enthused about investing in AMD hardware.

But I figured I would run a 10GB load of Linpack to see how the AS5 holds up compared to the other guys..

I am only running 2 intakes and an exhaust for case fans, and the stock FC140 fans.

 

[tabascosauz]

Needless to say that I am less than enthused about investing in AMD hardware.

View attachment 286321

If only I could also lower my room temp by 8°C you forever turn me green with envy

I am challenging myself to resist the upgrade bug and not to fix what ain't broke. Next round I will probably also take a break from Ryzen same as you.

However, I will be keeping my eye on Phoenix Point - should be a rather cost-effective way of getting my AM5 fix. Hope to do some wild memory and iGPU OC as with Cezanne I want some RDNA to play with.............just not in my main PC

 

[AusWolf]

No issues since disabling it, at least from what I can see. I've not rebooted in some time. I mean, I could reboot now and find out if it comes back up without having to play in UEFI again.

But I thought that a memory test like Windows Memory Diagnostics hammers memory with several types of tests to determine if any of the memory cells are bad.

Performed a warm boot, came back up with no issues. I then powered the system down, shut off power at the switch at the back of the power supply and held the power button on the front of the case down for some time just to make sure that I drained any and all capacitors on the motherboard. I then flipped the switch at the back of the power supply and booted back up. Again, no issues.

It seems that even with Memory Context Restore disabled the boot times are nowhere near as bad as it was under the old BIOS version. Perhaps keeping Memory Context Restore disabled is a good thing and shouldn't even be enabled if it only causes problems and doesn't cause boot length issues.

Memory Context Restore doesn't work on my board, either. Maybe leaving it off and forgetting that it even exists is the solution.

 

[freeagent]

Next round I will probably also take a break from Ryzen same as you.

I know this is a Ryzen thread, but I got this little guy here from a buddy

I need a third PC for my wife and youngest to play with

Which one they get I am not sure

ROG STRIX B660-I GAMING WIFI | ROG STRIX B660-I GAMING WIFI | Gaming Motherboards｜ROG - Republic of Gamers｜ROG Global (asus.com)