PS3 Project RSX Boost: Overclock your Retail PS3 RSX Speeds (ps3 cfw only)

I was able to figure out how to separate them. yeah, try that other link to see if something weird is happening.

upload_2025-1-12_7-11-13.png
 
great! not sure what's going on with that other one. I'll have to have @Mitsu™ take a look, I guess, at my oc task file. I don't see anything wrong with it.

upload_2025-1-12_7-18-36.png


as you can see the replace (first byte) is x06, which should be 300. I don't know if something is happening with the second value, since it's default.
 
Last edited by a moderator:
Yep looks like core freq is the culprit - did 4 test with 600/750, 650/800, artifacts at 700/800, settled on 650/850 and everything is pretty much stable. Funny how that extra 50mhz can make such a difference, but yeah you're right it behaves very similar to 2000 models. Not a dealbreaker tho coz I quite like the 21xx models over later revisions since it's better built (heatsink, wifi LEDs etc.)

Btw I love your yt vids, are you planning to test anymore games?
Yep, the 21** models are even tougher than the 25** !
And 650/850 is already an excellent OC !


About my YouTube channel, thanks for the kind words !
But no...
I don't plan on doing any new tests anytime soon. They take a lot of time (something I lack...), and honestly, I think I've already covered a good mix of games with different engines. It should give people a solid idea of what the RSX overclock can provide. This was my main objective.

On top of that... I tend to go all-in when I get into something (probably due to autism). I'm talking 24/7 focus to the point where it pushes everything else aside.

That's also why I turned off comments on my channel.
It wasn't that answering people bothered me. Far from it. I just needed to take a break.
That, and because I think this place is better for gathering informations. It's a great way to collect reports and share experiences, and there are people here who know way more than I do and can help answer questions better than I ever could.
View attachment 45222

as you can see the replace (first byte) is x06, which should be 300. I don't know if something is happening with the second value, since it's default.
separating them, I don't know if it affected anything. I didn't separate them with those previous two builds that worked, but I am using evilnat 4.91.3 beta-1 this time (if that has anything to do with it). it's possible the search hex is different somehow, negating the oc offsets, but that's just a guess:

https://1fichier.com/?l5dxlo3tljfldz62gamb
I don't see anything wrong in the picture you shared. Strange.
The 350/550 .PUP you built earlier worked perfectly, and it's also based on Evilnat 4.91.3 beta 1, so... yeah, I'm not sure x)

The 300/400 link you provided is working perfectly UwU
Nice !
I'm at 300/500 and wasn't sure about going below 500 MHz for VRAM.

By the way, at these frequencies, RetroArch doesn't boot. It just returns to the XMB. I tried launching Drakengard 3, but the console froze. I think it's a VRAM issue.
RIP-Felix or Nascar managed to run (Murder) Crysis 3 at 300 MHz core, but the VRAM was higher, IIRC.

PS1 and PS2 games still work like a charm, though !
This definitely confirms that the RSX is hardly used here, so yeah, it also confirms that overclocking won't improve the framerate in PS2 titles for those who still had doubts about it.

I think it could get really dangerous at/under 250 MHz for the core and under 400 MHz for VRAM.
 
Last edited:
I could try going under 250 on core to se if its possible but I don't know
Hmm, do you have an e3flasher and know how to use it ? Otherwise, I wouldn't try this, honestly.
250 on core might be good. But under...
It's just as, if not more, dangerous than running too high frequencies. At least when you brick a PS3 with an overclock that's too high, there's a small chance you can unbrick it by letting it cool down as much as possible.
But here, if it really can't boot... temperature won't matter.
So, you'd better have something to unbrick it on hand !
 
yeah, be careful. I think the reason it didn't work, the original, was that it was being read as "default" as I replicated the default portion of the oc, since the first value was the same: 300.
 
Back
Top