F5F Stay Refreshed Hardware Desktop CPU voltage configurations

CPU voltage configurations

CPU voltage configurations

Pages (3): Previous 1 2 3
I
ItzUtopia_PvP
Member
149
08-20-2022, 04:19 PM
#21
Using the PBO2 options in my signature setup. I'm running an NH-D15 with AM5 offset on a FD North (mesh) board, equipped with 4x140mm case fans. The fans are positioned on the rear section while leaving the center open. The ambient temperature during those tests wasn't unusual, as it was summer in my home which lacks central air (soon). I plan to test with a Cinebench R23. Concerning performance gains, the 7950x3D behaves oddly. Depending on the situation, I've observed up to 5.9GHz on CCD1, while CCD0 stays capped at 5.25GHz. Typically, with a complete 16c/32t setup you'll notice strong maximum boosts on both CCDs, but not with just 8c/32t. Even CCD1 underperforms since heat from the CCD0 seems to restrict its operation and TDP. I've conducted extensive game testing with various CCD configurations, even down to 2c/4t on CCD0, to check for higher sustained boosts—which it doesn't provide. If your CPU achieves greater boosts when PBO2 is disabled, then the offset likely isn't reliable. It's becoming common to accidentally set voltage limits on CPUs or GPUs during under-voltage attempts, as I've seen with the 5800x3D. That's why I always recommend confirming any PBO2 or under-voltage settings using Cinebench R23.
I
ItzUtopia_PvP
08-20-2022, 04:19 PM #21

Using the PBO2 options in my signature setup. I'm running an NH-D15 with AM5 offset on a FD North (mesh) board, equipped with 4x140mm case fans. The fans are positioned on the rear section while leaving the center open. The ambient temperature during those tests wasn't unusual, as it was summer in my home which lacks central air (soon). I plan to test with a Cinebench R23. Concerning performance gains, the 7950x3D behaves oddly. Depending on the situation, I've observed up to 5.9GHz on CCD1, while CCD0 stays capped at 5.25GHz. Typically, with a complete 16c/32t setup you'll notice strong maximum boosts on both CCDs, but not with just 8c/32t. Even CCD1 underperforms since heat from the CCD0 seems to restrict its operation and TDP. I've conducted extensive game testing with various CCD configurations, even down to 2c/4t on CCD0, to check for higher sustained boosts—which it doesn't provide. If your CPU achieves greater boosts when PBO2 is disabled, then the offset likely isn't reliable. It's becoming common to accidentally set voltage limits on CPUs or GPUs during under-voltage attempts, as I've seen with the 5800x3D. That's why I always recommend confirming any PBO2 or under-voltage settings using Cinebench R23.

S
snoepjez
Member
60
08-20-2022, 11:29 PM
#22
I rarely encounter 5.9ghz and I haven't seen a figure close to that before. My best results were around 5.7ghz, not in Cinebench. On Cinebench, my cores peak at about 5ghz with 8 cores and around 4.8 on the other 8. That seems normal, shouldn't my CPU be underperforming like that? It works whether PBO is enabled or not. I have no overclocking or undervolting.
S
snoepjez
08-20-2022, 11:29 PM #22

I rarely encounter 5.9ghz and I haven't seen a figure close to that before. My best results were around 5.7ghz, not in Cinebench. On Cinebench, my cores peak at about 5ghz with 8 cores and around 4.8 on the other 8. That seems normal, shouldn't my CPU be underperforming like that? It works whether PBO is enabled or not. I have no overclocking or undervolting.

D
DC70114
Junior Member
16
08-21-2022, 10:00 AM
#23
What scores are you seeing for the Cinebench R23? Those benchmarks came from different CCD configurations and run times. One example used a PBO2 setup with a 200MHz clock and -15mV curve, running in cache mode. The only consistent speed I observed was between 4.8 and 4.9GHz across all cores during the Cinebench test. The hardware info screenshots were taken specifically for that Cinebench session, so idle performance wasn't included.
D
DC70114
08-21-2022, 10:00 AM #23

What scores are you seeing for the Cinebench R23? Those benchmarks came from different CCD configurations and run times. One example used a PBO2 setup with a 200MHz clock and -15mV curve, running in cache mode. The only consistent speed I observed was between 4.8 and 4.9GHz across all cores during the Cinebench test. The hardware info screenshots were taken specifically for that Cinebench session, so idle performance wasn't included.

1
12conker26
Member
59
08-24-2022, 02:41 PM
#24
That's good to hear, I'm on the right track. I'm getting 36k, which fits perfectly with Cinebench like you said—definitely in the acceptable range.
1
12conker26
08-24-2022, 02:41 PM #24

That's good to hear, I'm on the right track. I'm getting 36k, which fits perfectly with Cinebench like you said—definitely in the acceptable range.

A
Ariadne111
Member
137
08-26-2022, 03:13 AM
#25
I haven't tested per core PBO2 yet, I think it might not be as strong on the 7950x3D because of its strict constraints. I haven't seen anyone run XOC on that board, though I've considered removing the firmware a few times. After delidding, the 7950x3D seems able to handle more than 5GHz across all cores, likely due to the additional heat spreader layers from solder and IHS being more restrictive than regular CPUs.
A
Ariadne111
08-26-2022, 03:13 AM #25

I haven't tested per core PBO2 yet, I think it might not be as strong on the 7950x3D because of its strict constraints. I haven't seen anyone run XOC on that board, though I've considered removing the firmware a few times. After delidding, the 7950x3D seems able to handle more than 5GHz across all cores, likely due to the additional heat spreader layers from solder and IHS being more restrictive than regular CPUs.

Pages (3): Previous 1 2 3