Anomaly detected in 5950/dark hero 8.
Anomaly detected in 5950/dark hero 8.
ok a little background 5950x, just to save ppl googletime its a 16 core 32 thread cpu. 3.4 base and 4.6 boost, least thats what the box said. idk i shipped my 5900 back to amazon for it lol rog dark hero viii / wifi 4090 oc / strix 4x16g corsair vengence rgb 18@3600, i think its 18-22-22-42? taking a wild guess, not looking at cpu-z right now. 1.35v theres more parts but i doubt thats related to this stuff tons of watercooling parts, custom loop, 1.25 gallons of water lets just say i took my closet and shoved it into a chassis pretty clueless when it comes to overclocking. so i spent at least 4 days tinkering with my pc and researching, restarting, writing down changes from prior stress tests, blah. 225/325/245 is stable, good cores are -10, ok -7 and crap are -3. held up for a bit then i upped the cpu throttle limit to 95c, now she is fine at 90.8c with a 1x scalar. so im kicking it playing forza all night and the thing poops on me. so i set the throttle to 85c. seldom do i ever hit 80 on TJmax but the cores get wild i suppose? idk your answers better than mine. so now she runs 4.9ghz +/- 200mhz varience across all cores at a cool 80c. huh? when the throttle was at 90 (default) or 95 it would get to 89c under load and creep on up to 90.8 and blackout. BUT my cores would tap down to 3.6ghz at the highest. but it was... stable? winrar usually blows up my cpu, nah it held fine. i was scared thats for sure but it worked lol why is it when my throttle limit was lowered, my cores consistently run faster and cooler? mind you the scalar is at 1x, idk what im doing lol even more strange, running off a fresh bios (aside from minor normal tweaks like DOCP/BAR) i get around 28300 and cooler temps...? with this setup i got 24800-25000. All on R23 btw why does it work better with a lower throttle point set? should the values for the per-core PBO be inverted? good cores being -3 not -10? clarification on the negative scalar would be beneficial. its kinda confusing with the wording/interpretation, a larger value when its set to negative? that statement alone negates itself! urgh, help please
Each processor varies. My 5900X runs at 260/170/200-25 all-core, plus 200. I notice boosts reaching 5150, a few hit 5100, others 5000, while most cap around 4800-4900. The best all-core performance is near 4750 with light settings, and heavier configurations drop to about 4500 at roughly 265w power.
Ok, so if it’s a good core use a smaller value? Say, -30 not -5 So-so cores, idk find a stable all core offset (do per core but -10 on all) crappy ones use a larger value, -3 not -12 Idk if I lost anything, it’s about 4.9 across all cores. Like I said, +\- 200 mhz per core. That’s with the throttle set back to 85 rather than stock 90. I had trouble before that cause it would crash so I bumped it up to 95. At that point I would get to 89-90 (stable) but 3.4-3.6ghz on all cores and the temps wouldn’t drop. single core benches though, 60c at 4.9-5.0 (my overclock setting is at zero, idc about another 200mhz) just confusing with the negating term. Not looking at bios “help” now but something like the greater the number the more voltage. So… if it’s negative, the higher the neg number it’s increasing? Get my logic here? Whatever the Mobo says for description it’s actually inverted due to - rather than +?
I think my aim is to reach each core quickly while keeping temperatures stable. Once that’s done, I’ll ease back a bit—around 3.2 or lower than the base. I don’t know which video I saw, but I followed that person’s setup and got 5.2 on all cores. Probably not every single one, maybe I wasn’t using hardware info then. Still, it worked out—like being in neutral but cranking up to around 8k RPM. You’re not delivering power to the road. Could just be me, but yeah.