DEPRESSED MAN RAMBLES ABOUT PC (Regrets/Issues/Overclocking)
101:06 Any Regrets?
03:18 Issues?
09:38 Cpu OC start (prev history)
24:35 Prime95 small fft 32 thread example (EDC 220 so we thermal throttled even all core)
28:36 Prime95 worst case scenario (8 thread, one CCX gets blasted over tiny surface area - Instant throttling)
34:35 Cinebench r20 multi example (single core was ~641 unoptimised)
36:46 Cinebench r23 multi example (single core was ~1629 unoptimised)
38:22 CPUZ bench (not a fan of it)
40:45 BIOS Settings PBO curve optimiser (-30 all except two best, -18 c4, -16 c6 in my case, EDC will affect stability)
49:37 Stock/prices in Australia
50:53 Hard drive bench 3 x mp600 in triple raid 0
57:15 Gpu section (were saving proper gpu talk for when we actually play a modern game RIP CYBERPUNK KURWA/Waiting for FarCry 6) Timespy unoptimised 19082
Benches highest weve seen so far
CBr20 119xx / 641 Single Thread (HUB review 10157 multicore stock)
CBr23 305xx? (Cant remember if we got 31k) / 1629 Single Thread
Timespy 19082 Unstable/Unoptimised (top 1% aus, cant compete against multigpu)
Firestrike 42345 (11th best in aus, accounting for duplicates, going against 6ghz intel w sli 1080s)
Long form ramble pc update. Kept putting it off but I feel ill never make a better version of it. Overclocking is far from optimal, its a ongoing work in progress Its a cycle of: Get excited about oc'ing, oc, get sick of it, use computer, repeat. More of a comedy skit then anything else :/
Cinebench r20 single thread: 641 (Stock can obviously max boost single thread/220 EDC which is more allcore/with things open/HUB review was 641)
Cinebench r23 single thread: 1625 (220 EDC which is more allcore/with things open)
Some parts got cut off in file transition.
We installed 33f bios lol to show many updates its had 32:45 When we crashed the gpu, we couldnt get it working again until we used the backup bios lol.
When I say 210 EDC is best, i meant best overall combination of single and multi core PBO.
The per core offsets are effected by the EDC value, so offsets stable at some EDC might not be stable at a different EDC. Also AGESA Updates/Bios version will affect per core offset stability as well.
Worst case scenario: So we forgot we were at 220 EDC, but one CCX getting blasted - So half the surface area, generating all the heat, which causes a heat transfer bottleneck I think.
3dmark seciton got cut off in chunk transition - 3dmark lets you put as many results as you want for the same config, so accounting for duplicate results we got 11th best in australia at the time (for firestrike) which I though was hilarious (going against 6ghz intel with sli 1080s). Timespy has better multi gpu scaling and more results so we were only top 1%.
timespy: 19082 (at unstable oc).