0 reactions
Intel 13900K/14900K Stability - Need more samples
I'm looking to see if others are able to recreate some issues I've found regarding the stability of the 14900k. Considering I've been experiencing memory related crashes (about 1 per week) but my memory has been tested to be stable I've started looking into stability issues regarding cache on the processor.
The stock cache multiplier shown in XTU is x50 or 5ghz. While testing 100% stock (with x50 multiplier shown in XTU) the cache appears to be running at a x45 multiplier instead (looking to see if this is just my system or if this is common). If you manually turn the cache frequency down to x49 the cache runs at 4.9ghz and if you bring it back up to x50 and hit apply it starts to run at 5ghz which should be the stock frequency according to XTU.
At this 5ghz cache frequency my system cannot run the Smallest FFT's (hits L1 and L2 Cache) in prime 95 at all and will immediately crash. The E-Cores appear to have issues first but this may come down to chip/chip variance. The Prime 95 blend test will also crash with just a -0.005v cache voltage offset on the P-core cache. This is right on the very edge of stability and in other workloads is likely an issue.
Looking to see if others can recreate this with their chips and also to see if manually turning the cache frequency down solves crashes for people running into issues. This seems to match up with Wendel's findings related to memory issues and is also a part of the chip that can degrade over time which seems to be related to this issue.
Also, if anyone experiencing crashing can also run event viewer (Win+R - Type in eventvwr) I want to see if the crashes you've had are related to error - c0000005. It is related to memory and is a part of the majority of the instability I've seen on my system and also shows up fairly often in searches about 13900k and 14900k instability.
#hardware #technology
source
I'm looking to see if others are able to recreate some issues I've found regarding the stability of the 14900k. Considering I've been experiencing memory related crashes (about 1 per week) but my memory has been tested to be stable I've started looking into stability issues regarding cache on the processor.
The stock cache multiplier shown in XTU is x50 or 5ghz. While testing 100% stock (with x50 multiplier shown in XTU) the cache appears to be running at a x45 multiplier instead (looking to see if this is just my system or if this is common). If you manually turn the cache frequency down to x49 the cache runs at 4.9ghz and if you bring it back up to x50 and hit apply it starts to run at 5ghz which should be the stock frequency according to XTU.
At this 5ghz cache frequency my system cannot run the Smallest FFT's (hits L1 and L2 Cache) in prime 95 at all and will immediately crash. The E-Cores appear to have issues first but this may come down to chip/chip variance. The Prime 95 blend test will also crash with just a -0.005v cache voltage offset on the P-core cache. This is right on the very edge of stability and in other workloads is likely an issue.
Looking to see if others can recreate this with their chips and also to see if manually turning the cache frequency down solves crashes for people running into issues. This seems to match up with Wendel's findings related to memory issues and is also a part of the chip that can degrade over time which seems to be related to this issue.
Also, if anyone experiencing crashing can also run event viewer (Win+R - Type in eventvwr) I want to see if the crashes you've had are related to error - c0000005. It is related to memory and is a part of the majority of the instability I've seen on my system and also shows up fairly often in searches about 13900k and 14900k instability.
#hardware #technology
source
| 31,000 mlx | |
| 45 | |
| 0 |