PS3 Project RSX Boost: Overclock your Retail PS3 RSX Speeds (ps3 cfw only)
Go to cfw tools > dump tools > Dump LV1 and send me the file via DM.
trick ster
Member
@LuanTeles could you send me a link for 1000mhz overclock are help me in anyway to get a 1000mhz Update file please?
Naxil
Member
Iam not dev, just wondering.
Rules u need to patch cfw for change rsx mhz value.
You have two slot for cfw right?
Slot 1 and slot 2
Vanilla slot 1 patch slot 2, reboot and point the boot on slot 2.. fails? So you cant update a flag, then next reboot ps3 load slot 1. If the flag is in "working" status continue to use slot2. Is possible? Like the way you change resolution on pc and it ask you it is right? If you put yes you stay on new resolution. Same logic but with reboot and flag in the middle...
Tanzu15
Member
im guessing 1 month apart might not matter too much. 750/950 should work. but please test many many games. cause some games might not bug out but others might. keep a close eye for artifacts. if after many games and hours its stable, congrats. you got a solid huge overclock.nvm i got the month mixed up with my original model. mine is october 2010
hey i sent you the new lv1 dump from my other ps3.
whats the model number? and check the bottom of ps3 for manufacture date.
Last edited by a moderator:
Cratus666
Member
Hey guys another update. I am currently running 750/1000 still no artifacts, I am playing shift 2 unleashed as we speak, Further updates will come
Edit: Shift 2 has began to show artifacts and same with oblivion but had no issues with bioshock infinite
CECH 2504B Can run 750/950 and 700/1000 with no issues
Edit: Shift 2 has began to show artifacts and same with oblivion but had no issues with bioshock infinite
CECH 2504B Can run 750/950 and 700/1000 with no issues
Last edited:
MistAndYork
Forum Noob
MGS4 have been good to me when checking for artifacts, just at the beginning of the game when it tries to hit 60fps. Other games that ive tried that 100% push the hardware harder but didnt produce artifacts were crysis 3 and Castlevania lords of shadow.
cha0shacker
Member
The first Crysis game was the only game on my system to consistently show instabilities.
Tanzu15
Member
Not bad at all. Though I hope you test a large number of games, cause it might take certain games to see the instability. It's why I test new games every day. So far I'm blessed with 750/950.
Thimblewad
Member
Hey guys!
Sorry, I've been away for a couple of days. Very nice to see quite some new faces testing and trying out new things, good job, that's why we're all here!
Sorry, I've been away for a couple of days. Very nice to see quite some new faces testing and trying out new things, good job, that's why we're all here!
Tanzu15
Member
No problem bro. It's been hella fun giving data has been fun. I fully understand now, which ps3s can reach what frequencies.Hey guys!
Sorry, I've been away for a couple of days. Very nice to see quite some new faces testing and trying out new things, good job, that's why we're all here!
@LuanTeles have you tested 850/1000 yet?
Last edited by a moderator:
JoeSapecudo
Member
Looking forward to testing 850/1000 on mine, and from the information already collected, apparently the best (RSX) for overclocking are from the 25xx consoles
Correct me if I'm wrong.
Correct me if I'm wrong.
Before I start interjecting here I'll preface this with I've read the last 27 pages of this thread and 10 or so from the start of it, skipped some pages as nothing of real consequence was being discussed. What I'm about to say doesn't just come from someone chiming in randomly I'm the guy that figured out Polaris and wrote THIS OC guide\article so if you want someone with extremely in-depth hardware knowledge to accurately figure out the RSX, then I can help with that. I'll start with some general stuff I've made mental notes of and work my way to more specific things.
@LuanTeles I believe who is trying to get some sort of semblance for a standardised testing method, it is imperative that everyone test with the same methodology to get this RSX testing right. Namely, start with OCing the RSX memory in 25 or 50MHz steps to identify your maximum clock, then from there to account for different games pushing the hardware in different ways you should back off 50MHz to ensure and maintain absolute stability. Once this is done memory clocks should be reverted to stock and the process repeated for the GPU. Eg; if individually you have stable clocks of 750\800 you would be wise to set combined clock values to 700\750. Doing this it would also be easy and clear to measure where the biggest bottleneck really is in different games - bandwidth or raw rasterization performance. I won't start talking about the IMC at this point to try and not start confusing people. In terms of testing Crysis and MGS4 should prove quite reliable for consistent results as will Lego (yes lego) Jurassic World because while it won't push the hardware as much as other games it has terrible frame pacing on PS3 so will stress the hardware in a different way.
@LuanTeles To add some explanation about RSX memory clocks and why it seems to vary so much it has nothing to really do with the RSX itself but the GDDR, it's not just a dice roll of the silicon lottery or trace length in this situation. On early PS3s (up to end 2009 I'd hazard an educated estimate) the RSX uses Samsung K4J52324QE-RC14, it is specified to operate at 700MHz (1.4GHz effective), 1.8v. So there we have the absolute maximum safe minimum value for all PS3s an OC firmware can be set to, beyond that silicon lottery plays more of a part, but not entirely, and here's why. For people hitting 800-900MHz+ but not 1GHz+ you likely have the slightly higher specified -BC12 memory, and for those hitting 1GHz without problems consider yourself luck, there's a good chance you have either -BJ11 (specified for 900MHz) or -BJ1A (specified for 1GHz). Here is where I'd usually start talking about GDDR memory timings and how to optimise them but as I doubt anyone has found the timing straps for it yet I won't touch on this.. for now.
Now, about the RSX itself and core clock frequencies, on 90nm lithography chips you should consider 550-600MHz the absolute maximum you push them and even that is a risk not only due to the archaic 90nm node but also because of the well documented heat and sub-standard materials issues. I won't go into details about that but if he feels like it I'm sure @RIP-Felix can regale you all with his tales on the subject or link you all to some videos.
For the 65nm lithography RSX, assuming you aren't unlucky to have one of the early slims with "hybrid" RSX stock using bad underfill you should consider 650MHz the absolute maximum for completely stable operation. I will only touch on this lightly so as to keep things as easy to follow as possible, each process node, regardless of if it's the ancient 90nm or cutting edge 5nm and below has what is known as critical points, this is the scale at which frequency scales linearly with voltage, it would be wise to remember this so as not to over push the RSX and end up with an oversized paperweight where the RSX can't boot the attempted frequency due to a lack of voltage because it's fallen off the scale.
The 40nm RSX is where things get interesting, this node had the most development and matured faster so taking into consideration less than ideal ASIC quality (<59%) which would lead to more leakage and considering the packages with ideal ASIC quality (69-75%, a balance between package leakage, and required voltage for a desired core clock) as well as pretty much perfect ASIC quality (90%+, which ironically isn't actually ideal for high core frequencies) then you can put the absolute maximum core frequency at anywhere between 700MHz - 850MHz.
The 28nm RSX should be able to do anything between 800MHz-1GHz+, if super slims can ever be cracked.
These things are crucial to know because if you are trying to create a "one size fits all" OC firmware then you should absolutely leave the core clock frequency at 600MHz. If we forget about fat model PS3s (and we should for their own sake) then a safe, absolutely stable frequency for all slims should be 650MHz. If individual owners choose to go above that frequency it's their own arse, but don't forget a 650MHz core clock is a 30% overclock over the stock 500MHz, if you can't hit a target 30FPS with an OC like that hardware is not the problem - shitty game coding and optimisation is.
@Succ I have a suggestion, open the PS3s PSU. Inside you should find two adjustable (with a screwdriver) potentiometers (aka "pots"). Adjusting these you can make the PSU output more voltage over the rails, not recommended for long term use but if your PS3 just requires some extra juice to boot it up and flash it back to some more conservative clocks, this should do it.
Probably 650/750 max for 20XX models not just because of the 65nm node but also because NEC/TOKINS, they are a weak link as it is.
@cha0shacker create a mega.nz account 50GB storage with even a free one.
@cha0shacker there might be a way to control it in software actually. If you know where the VRM signals are coded in the firmware you can skew the signal to force a higher voltage and as far as the system is concerned it's still pushing default voltages. Old trick to compensate for vdroop, just be careful with what skewed values to push to the VRM so you don't blow it up (literally).
If anyone has any questions just tag me or shoot a DM my way. I have a couple PS3 slims (2103 & 2503) I'll target high but within technical specifications for when a final Evilnat 4.91.x CFW is released, probably 650\700 as my baseline with a goal of seeing if 700/850 is stable.
@LuanTeles I believe who is trying to get some sort of semblance for a standardised testing method, it is imperative that everyone test with the same methodology to get this RSX testing right. Namely, start with OCing the RSX memory in 25 or 50MHz steps to identify your maximum clock, then from there to account for different games pushing the hardware in different ways you should back off 50MHz to ensure and maintain absolute stability. Once this is done memory clocks should be reverted to stock and the process repeated for the GPU. Eg; if individually you have stable clocks of 750\800 you would be wise to set combined clock values to 700\750. Doing this it would also be easy and clear to measure where the biggest bottleneck really is in different games - bandwidth or raw rasterization performance. I won't start talking about the IMC at this point to try and not start confusing people. In terms of testing Crysis and MGS4 should prove quite reliable for consistent results as will Lego (yes lego) Jurassic World because while it won't push the hardware as much as other games it has terrible frame pacing on PS3 so will stress the hardware in a different way.
@LuanTeles To add some explanation about RSX memory clocks and why it seems to vary so much it has nothing to really do with the RSX itself but the GDDR, it's not just a dice roll of the silicon lottery or trace length in this situation. On early PS3s (up to end 2009 I'd hazard an educated estimate) the RSX uses Samsung K4J52324QE-RC14, it is specified to operate at 700MHz (1.4GHz effective), 1.8v. So there we have the absolute maximum safe minimum value for all PS3s an OC firmware can be set to, beyond that silicon lottery plays more of a part, but not entirely, and here's why. For people hitting 800-900MHz+ but not 1GHz+ you likely have the slightly higher specified -BC12 memory, and for those hitting 1GHz without problems consider yourself luck, there's a good chance you have either -BJ11 (specified for 900MHz) or -BJ1A (specified for 1GHz). Here is where I'd usually start talking about GDDR memory timings and how to optimise them but as I doubt anyone has found the timing straps for it yet I won't touch on this.. for now.
Now, about the RSX itself and core clock frequencies, on 90nm lithography chips you should consider 550-600MHz the absolute maximum you push them and even that is a risk not only due to the archaic 90nm node but also because of the well documented heat and sub-standard materials issues. I won't go into details about that but if he feels like it I'm sure @RIP-Felix can regale you all with his tales on the subject or link you all to some videos.
For the 65nm lithography RSX, assuming you aren't unlucky to have one of the early slims with "hybrid" RSX stock using bad underfill you should consider 650MHz the absolute maximum for completely stable operation. I will only touch on this lightly so as to keep things as easy to follow as possible, each process node, regardless of if it's the ancient 90nm or cutting edge 5nm and below has what is known as critical points, this is the scale at which frequency scales linearly with voltage, it would be wise to remember this so as not to over push the RSX and end up with an oversized paperweight where the RSX can't boot the attempted frequency due to a lack of voltage because it's fallen off the scale.
The 40nm RSX is where things get interesting, this node had the most development and matured faster so taking into consideration less than ideal ASIC quality (<59%) which would lead to more leakage and considering the packages with ideal ASIC quality (69-75%, a balance between package leakage, and required voltage for a desired core clock) as well as pretty much perfect ASIC quality (90%+, which ironically isn't actually ideal for high core frequencies) then you can put the absolute maximum core frequency at anywhere between 700MHz - 850MHz.
The 28nm RSX should be able to do anything between 800MHz-1GHz+, if super slims can ever be cracked.
These things are crucial to know because if you are trying to create a "one size fits all" OC firmware then you should absolutely leave the core clock frequency at 600MHz. If we forget about fat model PS3s (and we should for their own sake) then a safe, absolutely stable frequency for all slims should be 650MHz. If individual owners choose to go above that frequency it's their own arse, but don't forget a 650MHz core clock is a 30% overclock over the stock 500MHz, if you can't hit a target 30FPS with an OC like that hardware is not the problem - shitty game coding and optimisation is.
@Succ I have a suggestion, open the PS3s PSU. Inside you should find two adjustable (with a screwdriver) potentiometers (aka "pots"). Adjusting these you can make the PSU output more voltage over the rails, not recommended for long term use but if your PS3 just requires some extra juice to boot it up and flash it back to some more conservative clocks, this should do it.
Probably 650/750 max for 20XX models not just because of the 65nm node but also because NEC/TOKINS, they are a weak link as it is.
@cha0shacker create a mega.nz account
50GB storage with even a free one.@cha0shacker there might be a way to control it in software actually. If you know where the VRM signals are coded in the firmware you can skew the signal to force a higher voltage and as far as the system is concerned it's still pushing default voltages. Old trick to compensate for vdroop, just be careful with what skewed values to push to the VRM so you don't blow it up (literally).
If anyone has any questions just tag me or shoot a DM my way. I have a couple PS3 slims (2103 & 2503) I'll target high but within technical specifications for when a final Evilnat 4.91.x CFW is released, probably 650\700 as my baseline with a goal of seeing if 700/850 is stable.
Tanzu15
Member
Bro you don't want to try that 850/1000. That's an insane OC that can cause a brick. 25xx consoles cannot do that stable. Hell the risk of brick is high.
This was pretty educational. But I knew a few things. Luckily I've fully understand what ps3s can handle depending on the model. And what frequency ranges.Before I start interjecting here I'll preface this with I've read the last 27 pages of this thread and 10 or so from the start of it, skipped some pages as nothing of real consequence was being discussed. What I'm about to say doesn't just come from someone chiming in randomly I'm the guy that figured out Polaris and wrote THIS OC guide\article so if you want someone with extremely in-depth hardware knowledge to accurately figure out the RSX, then I can help with that. I'll start with some general stuff I've made mental notes of and work my way to more specific things.
@LuanTeles I believe who is trying to get some sort of semblance for a standardised testing method, it is imperative that everyone test with the same methodology to get this RSX testing right. Namely, start with OCing the RSX memory in 25 or 50MHz steps to identify your maximum clock, then from there to account for different games pushing the hardware in different ways you should back off 50MHz to ensure and maintain absolute stability. Once this is done memory clocks should be reverted to stock and the process repeated for the GPU. Eg; if individually you have stable clocks of 750\800 you would be wise to set combined clock values to 700\750. Doing this it would also be easy and clear to measure where the biggest bottleneck really is in different games - bandwidth or raw rasterization performance. I won't start talking about the IMC at this point to try and not start confusing people. In terms of testing Crysis and MGS4 should prove quite reliable for consistent results as will Lego (yes lego) Jurassic World because while it won't push the hardware as much as other games it has terrible frame pacing on PS3 so will stress the hardware in a different way.
@LuanTeles To add some explanation about RSX memory clocks and why it seems to vary so much it has nothing to really do with the RSX itself but the GDDR, it's not just a dice roll of the silicon lottery or trace length in this situation. On early PS3s (up to end 2009 I'd hazard an educated estimate) the RSX uses Samsung K4J52324QE-RC14, it is specified to operate at 700MHz (1.4GHz effective), 1.8v. So there we have the absolute maximum safe minimum value for all PS3s an OC firmware can be set to, beyond that silicon lottery plays more of a part, but not entirely, and here's why. For people hitting 800-900MHz+ but not 1GHz+ you likely have the slightly higher specified -BC12 memory, and for those hitting 1GHz without problems consider yourself luck, there's a good chance you have either -BJ11 (specified for 900MHz) or -BJ1A (specified for 1GHz). Here is where I'd usually start talking about GDDR memory timings and how to optimise them but as I doubt anyone has found the timing straps for it yet I won't touch on this.. for now.
Now, about the RSX itself and core clock frequencies, on 90nm lithography chips you should consider 550-600MHz the absolute maximum you push them and even that is a risk not only due to the archaic 90nm node but also because of the well documented heat and sub-standard materials issues. I won't go into details about that but if he feels like it I'm sure @RIP-Felix can regale you all with his tales on the subject or link you all to some videos.
For the 65nm lithography RSX, assuming you aren't unlucky to have one of the early slims with "hybrid" RSX stock using bad underfill you should consider 650MHz the absolute maximum for completely stable operation. I will only touch on this lightly so as to keep things as easy to follow as possible, each process node, regardless of if it's the ancient 90nm or cutting edge 5nm and below has what is known as critical points, this is the scale at which frequency scales linearly with voltage, it would be wise to remember this so as not to over push the RSX and end up with an oversized paperweight where the RSX can't boot the attempted frequency due to a lack of voltage because it's fallen off the scale.
The 40nm RSX is where things get interesting, this node had the most development and matured faster so taking into consideration less than ideal ASIC quality (<59%) which would lead to more leakage and considering the packages with ideal ASIC quality (69-75%, a balance between package leakage, and required voltage for a desired core clock) as well as pretty much perfect ASIC quality (90%+, which ironically isn't actually ideal for high core frequencies) then you can put the absolute maximum core frequency at anywhere between 700MHz - 850MHz.
The 28nm RSX should be able to do anything between 800MHz-1GHz+, if super slims can ever be cracked.
These things are crucial to know because if you are trying to create a "one size fits all" OC firmware then you should absolutely leave the core clock frequency at 600MHz. If we forget about fat model PS3s (and we should for their own sake) then a safe, absolutely stable frequency for all slims should be 650MHz. If individual owners choose to go above that frequency it's their own arse, but don't forget a 650MHz core clock is a 30% overclock over the stock 500MHz, if you can't hit a target 30FPS with an OC like that hardware is not the problem - shitty game coding and optimisation is.
@Succ I have a suggestion, open the PS3s PSU. Inside you should find two adjustable (with a screwdriver) potentiometers (aka "pots"). Adjusting these you can make the PSU output more voltage over the rails, not recommended for long term use but if your PS3 just requires some extra juice to boot it up and flash it back to some more conservative clocks, this should do it.
Probably 650/750 max for 20XX models not just because of the 65nm node but also because NEC/TOKINS, they are a weak link as it is.
@cha0shacker create a mega.nz account
@cha0shacker there might be a way to control it in software actually. If you know where the VRM signals are coded in the firmware you can skew the signal to force a higher voltage and as far as the system is concerned it's still pushing default voltages. Old trick to compensate for vdroop, just be careful with what skewed values to push to the VRM so you don't blow it up (literally).
If anyone has any questions just tag me or shoot a DM my way. I have a couple PS3 slims (2103 & 2503) I'll target high but within technical specifications for when a final Evilnat 4.91.x CFW is released, probably 650\700 as my baseline with a goal of seeing if 700/850 is stable.
Last edited by a moderator:
GuilloteTesla
Moderator
Friendly reminder: avoid post multiple messages one after the other as it turns difficult to follow and reply messages. Use the edit function to add more replies to the original message.
Thank you.
Thank you.
trick ster
Member
where can i get my hands on diffrent overclock versions. im at 650mhz but need to go higher. how do i do it? i tried the custom firmware maker put i get patch tool error
RIP-Felix
Senior Member
AFAIK the 65nm RSX has good underfill. I tested a 2982 (1st revision) and it wasnt the the same low Tg stuff the 90nm has. So unless you have information I've not seen the 65nm are all in the clear as far as defects go. They are tanks.
I would like to get your thoughts on the effect of ripple/noise with higher clocks. I suspect that increasing them increases noise and ripple that makes any OC less stable, especially near the limit. FBVDDQ (VRAM voltage) on a COK-00X MB has 2 unpopulated 7343 pads for polymer bulk filtering caps (tantalum). By default there's 1. I have been wondering if populating those would improve long term reliability of the RSX. OTOH, populating those "may" allow you to clock higher too. A 90nm RSX it's not exactly reccomended, given it's thermal limitation (70c) and related reliability concerns. But for a frankie with a 65 or 40nm, the heatsink, VRM, and additional filtering pads have the potential to really push the limits.
Combine that with @vyktormvmpay25's VRAM milling process for removing and replacing the VRAM on the RSX package. We could concievably replace a 40nm's VRAM with the higest specification compatible.
Would be interested to see what that puppy could manage!
Last edited by a moderator:
JoeSapecudo
Member
Similar threads
-
PS3 [Guide] Sourcing the right PS3 for the best Overclocking performance
- Started by M4N14C
- Replies: 1
-
PS3 {CFW Release} 4.92.2 Evilnat Custom Firmware (w/ Cobra v8.5) - Includes: CEX/PEX/D-PEX) + noBD/BT- Started by Evilnat
- Replies: 1
-
-
PS3 PS3HEN w/ Linux & Overclocking Support? BadHTAB A hardware-software based hypervisor (lv1) exploit- Started by STLcardsWS
- Replies: 52