thekingsnigthmare
Member
So I should straight install the level of overclock I want and everything will be fine.
PS3 Project RSX Boost: Overclock your Retail PS3 RSX Speeds (ps3 cfw only)
Mitsu™
Senior Member
If you want to install 700/900 on a 25XX, then yes, it's 100% safe.
Starting from 800MHz core, there are no guarantees, especially without knowing your RSX model.
Same for VRAM, don't jump directly to 1000MHz.
thekingsnigthmare
Member
RIP-Felix
Senior Member
Haven't been keeping up with this thread, but thought I should post a few things.
1st. Higher frequencies lead to more work, which means more current, which means more heat. I tracked this in my own tests which I've not had a chance to compile and post yet. But there is a quite obvious trendline in which heat scales with CORE frquency increases. VRAM OTOH doesn't contribute to the heat until it reaches a stability threshold, then skyrockets. I'm not sure what to make of this yet. Might be thermal transfer, might be architectural...IDK.
***EDIT***
I just realized why this happened. When I first started testing I had the testbench console laying flat against the table, which restricted airflow to the Fan. There were about 4 early measurements like that before I started propping it up so more air can get in unhindered. I'm glad I recorded video so I could review it after noticing a strange trend in the data like that. There were several odd data points where the temps were significan;tly higher without seeming explanation. They all has that in common. It was not getting enough air and heating up more because of it. It wasn't that the VRAM was getting hot nearing the highest frequency.
I'm redoing those measurements now so that the data is accurate and comparable.*****
2nd. The extra work being done will increase the destructive effects of electromigration. The forces applied on the actual traces and transistors inside the die due to the increased activity of electrons passing through them...or more accurately, the electromagnetic fields...causes a breakdown in the circuit. The copper can deform and break, leadi g to an open fault. The dielectric walls that separate them can be broken through and traces short. This process takes time and when we increase frequencies and/or voltage, this time is shortened.
So 100% fact, OVERCLOCKING WILL SHORTEN THE LIFE OF YOUR PS3.
1st. Higher frequencies lead to more work, which means more current, which means more heat. I tracked this in my own tests which I've not had a chance to compile and post yet. But there is a quite obvious trendline in which heat scales with CORE frquency increases. VRAM OTOH doesn't contribute to the heat until it reaches a stability threshold, then skyrockets. I'm not sure what to make of this yet. Might be thermal transfer, might be architectural...IDK.
***EDIT***
I just realized why this happened. When I first started testing I had the testbench console laying flat against the table, which restricted airflow to the Fan. There were about 4 early measurements like that before I started propping it up so more air can get in unhindered. I'm glad I recorded video so I could review it after noticing a strange trend in the data like that. There were several odd data points where the temps were significan;tly higher without seeming explanation. They all has that in common. It was not getting enough air and heating up more because of it. It wasn't that the VRAM was getting hot nearing the highest frequency.
I'm redoing those measurements now so that the data is accurate and comparable.*****
2nd. The extra work being done will increase the destructive effects of electromigration. The forces applied on the actual traces and transistors inside the die due to the increased activity of electrons passing through them...or more accurately, the electromagnetic fields...causes a breakdown in the circuit. The copper can deform and break, leadi g to an open fault. The dielectric walls that separate them can be broken through and traces short. This process takes time and when we increase frequencies and/or voltage, this time is shortened.
So 100% fact, OVERCLOCKING WILL SHORTEN THE LIFE OF YOUR PS3.
Last edited:
Mitsu™
Senior Member
Thanks for this ! It's a precious intervention
Indeed, this is 100% factual, just rules of physics and thermodynamics.
Claiming the contrary is pure misinformation, a PS3 is no exception to these laws.
The myth of the "risk-free OC" is firmly anchored, unfortunately.
The higher you push your GPU, the more electromigration is exacerbated.
That's why I recommend to not OC the RSX higher than 700/800-850, even for a 25XX.
Ketxxx is right, 650/800 is already enough for a good amount of games (it's the recommended OC for a 20XX).
Considering the diminishing returns process, higher frequencies aren't worth the risks.
And yes, it's (diminishing returns) also an inevitable process that affects any GPU. Especially without voltage modification, this is something that we run into relatively quickly. The RSX is nothing special.
(That said, a voltmod would worsen electromigration + heat management seems problematic. Still an interesting experiment though
)
Last edited:
Tanzu15
Member
So from 30 years down to 25. Won't matter much. As time goes by the PS3s will inevitably die anyways, or be so old, they will be irrelevant. Heat is the reason for its death mostly. It's why people really shouldn't run their PS3s with a fan speed of 25% to 30%. My PS3s are always at 40 to 45% fan speed. And I never go above 65c. We also don't play on the PS3 for 8 to 16 hours a day. These Overclocks will still out live us. PS3s in general will out live us. Worrying about electronics dying is pointless. What's most important is to keep it clean, and cool. None of us have Time or desire to play on these consoles 8 to 12 hours a day for years. That time has long passed when we were kids or teenagers.
RIP-Felix
Senior Member
Best not to pull numbers out of your butt.
I have been conducting these tests on a well worn console with 1125 days of use on it. I have seen some slims make it to 1500, but not many, Most of the time they don't make it that long. I can't say it was difinitvly electromigration that took them out, but I am seeing the caps start to wear out on this console. And I'm getting a periodic colored noise frame, a flash of colored static, sometimes the entire frame, sometimes half the screen, that I'm not entirely sure isn't the acumulated wear an tear on this RSX. Now, it could be my test setup. Capture card. The 25-foot HDMI cable to my test bench. The HDMI transmitter or DVE...etc. So yeah, I'm not drawing any conclusions about how long the RSX can last and if my console actually has electromigration damage showing up in any tangible manner.
I'm just saying, this console was manufactured in 2010 and 14 years later is showing it's age. I highly doubt it'll make 25 or 30, without at least a new set of VRM, caps, and another RSX.
As for worring about electronics dying, you have a point. You shouldn't worry about them dying. That's not to say they'll outlive us. The WONT, unless you like to ride motorcycles. Then it'll probably outlive YOU...but not the rest of us. The reason people latch onto this false narrative about their beloved electronics living forever is because they don't want to accept the reality that electronics are perishables.
Capacitors for example have a shelf life that's 25-30 years even if they are not used. Shorter when used. The clock battery can corrode and leak onto the motherboard, destroying traces...SEAL IN IT'S ORIGINAL BOX!
It's comforting to think electronics aren't perishable like the cabbage in your fridge. But it's a mental crutch those in denial use for avoiding the painful truth. The sooner you realize and accept the facts, the sooner you can stop worring about the inevitable and just enjoy it while it lasts.
I have been conducting these tests on a well worn console with 1125 days of use on it. I have seen some slims make it to 1500, but not many, Most of the time they don't make it that long. I can't say it was difinitvly electromigration that took them out, but I am seeing the caps start to wear out on this console. And I'm getting a periodic colored noise frame, a flash of colored static, sometimes the entire frame, sometimes half the screen, that I'm not entirely sure isn't the acumulated wear an tear on this RSX. Now, it could be my test setup. Capture card. The 25-foot HDMI cable to my test bench. The HDMI transmitter or DVE...etc. So yeah, I'm not drawing any conclusions about how long the RSX can last and if my console actually has electromigration damage showing up in any tangible manner.
I'm just saying, this console was manufactured in 2010 and 14 years later is showing it's age. I highly doubt it'll make 25 or 30, without at least a new set of VRM, caps, and another RSX.
As for worring about electronics dying, you have a point. You shouldn't worry about them dying. That's not to say they'll outlive us. The WONT, unless you like to ride motorcycles. Then it'll probably outlive YOU...but not the rest of us. The reason people latch onto this false narrative about their beloved electronics living forever is because they don't want to accept the reality that electronics are perishables.
Capacitors for example have a shelf life that's 25-30 years even if they are not used. Shorter when used. The clock battery can corrode and leak onto the motherboard, destroying traces...SEAL IN IT'S ORIGINAL BOX!
It's comforting to think electronics aren't perishable like the cabbage in your fridge. But it's a mental crutch those in denial use for avoiding the painful truth. The sooner you realize and accept the facts, the sooner you can stop worring about the inevitable and just enjoy it while it lasts.
Last edited:
Tanzu15
Member
1500 days is already plenty of service time. That's insane amount of days. That right there proves, these things will outlive their usefulness. And now imagine someone having 2 as back ups, it's a none issue. Never was, never will be. These PS3s don't get abused like they used to anymore. Once all the online servers are shut down along with the store, it gets worst for them. 1500 days is insane. Plus PS3 die randomly out of no where anyways. Even at stock. Just taking one appart kills them randomly. Them dying much faster due to taking them appart will be something that happens way more often than Overclocks. None of my 11 PS3s have anywhere near 1500 days. I know someone else that has a 1500 or so console and still works. These things are very very very long lasting. My highest day count on one of my slims is around 350 days. I have 2 that OC high at 120 days and the one that does 950 core is at a whopping 350 ish days. Worrying about PS3 longevity is literally pointless. They aren't expensive, are readily available, and last very very long, once again as long as you actually maintain them clean and cool. Worrying about OC longevity is a none issue. At least not on 40nm chips. Maybe 90 or 65nm, and even then, not really. Overclocked consoles or GPUs, by the time they are dying, they will have been heavily used and no longer of use. Plus I'd say in a few more years, RPCS3 will be so perfect PS3s are useless. Though I personally always prefer native PS3 cause I have emotional attachment to PS3 in general. But the best experience will be the emulator, which will let you use the xmb and everything like a native PS3. And will have online built in. Edit:. Most PS3 no way in hell made it to 1500 days of use. That's a crazy number. And 1500 still shows the PS3 is alive and kicking. Still not dead, so it could likely squeeze out more years. Also the 30 to 25 was just an example. PS3s sh!t the bed as soon as 50 days. Weather it be a manufacture defect or something else died, PS3s die all the time, for no reason. One day it's working the next day it ain't. And almost no local shops know sh!t about PS3 repairs, and people that know how to repair scalp the repair likes it some sort of rocket science, so most normal people would just throw it out, and buy another one for cheap on eBay, Facebook market place or offer up. I've bought 3 slims on fb and offer up for 20 a pop. Great condition, one being 65 day up time.
RIP-Felix
Senior Member
Interesting result worth a separate post. Before my highest stable OC was 700/950 (CECH 2501a). At 750 core it had slight artifacting. I did all those voltage hikes to reveal the capacitors were aged, but not failed yet and added 3x AlPol to the RSX side. Which sured up the filter and did help to stabilize one of those overclocks. But I just now realized that 750 core is stable without a voltage hike as well. The artifacting that I was seeing before isn't here. So now, my highest stable OC is 750/950...I assume I can't go any higher.
Last edited:
Tanzu15
Member
This is your Frankie right? Or slim?
Raidmax
Member
Just for the fun of it, I've run a 750/975 PUP (from @Mitsu™ 's Mediafire repository) on a spare 2503A 0C slim that I'm not sure if I should torture anymore or just let the mobo die and swap another one (the other one I have has an less desirable RSX... CXD5300A1GB. And a missing HDMI socket but that's for another day - at least that one still runs with no issues.) in place. RSX model in this 0C's case is a CXD5300CGB.
Basically it started GLOD-ing (w/ no codes - I checked the logs in both instances, prior to GLOD and after it) after taking apart. I knew something went bad as pressure test would pass, so for the time being a small copper penny sits inbetween the RSX and its respective heatsink until I swap a whole mobo in. (either the A1GB I extracted from the scraps of a 0D unit and keep it as a spare stock slim or any board I can find with a better RSX)
Tried to "kill" it off with a straight 750/975 OC off the bat And... I'm not sure if I should laugh or just put it out of its misery because what happened once the update was done was quite funny to me.
It runs, no artefacts, but good god trying to run ANYTHING on it halts to a crawl, except the XMB waves.
I'm not sure if the HDD is at fault here since the stock speed CFW it had prior (4.90) ran at normal speeds, no slowdowns - you could browse the XMB without it slowing down to a crawl. What a shame... I'd have run Crysis HD just to see how it fares in its actual "life support" state w/ the 750/975 CFW.
Anyways, this was my first look into one of the OC firmwares. I saved a few just in case I land a good slim locally and... one of the 900 CFW will serve as the CGB slim's funeral once I find a newer 0D/1A unit that can do CFW and hasn't been as abused prior as this one.
I also have not tried phats as of yet, mainly due to time constraint. Nothing interesting going thru my sig, except for a CECHJ04 (which I forgot to edit the list to show up in there) that has Chemicon stock caps instead of NEC, that I suppose I can try to push a lil' bit.
In the end, I'm calling this a night, it's 3:16 AM here already. I might come back tomorrow if I have any luck with the J04, it needs its fill of games anyways.
Basically it started GLOD-ing (w/ no codes - I checked the logs in both instances, prior to GLOD and after it) after taking apart. I knew something went bad as pressure test would pass, so for the time being a small copper penny sits inbetween the RSX and its respective heatsink until I swap a whole mobo in. (either the A1GB I extracted from the scraps of a 0D unit and keep it as a spare stock slim or any board I can find with a better RSX)
Tried to "kill" it off with a straight 750/975 OC off the bat And... I'm not sure if I should laugh or just put it out of its misery because what happened once the update was done was quite funny to me.
It runs, no artefacts, but good god trying to run ANYTHING on it halts to a crawl, except the XMB waves.
I'm not sure if the HDD is at fault here since the stock speed CFW it had prior (4.90) ran at normal speeds, no slowdowns - you could browse the XMB without it slowing down to a crawl. What a shame... I'd have run Crysis HD just to see how it fares in its actual "life support" state w/ the 750/975 CFW.
Anyways, this was my first look into one of the OC firmwares. I saved a few just in case I land a good slim locally and... one of the 900 CFW will serve as the CGB slim's funeral once I find a newer 0D/1A unit that can do CFW and hasn't been as abused prior as this one.
I also have not tried phats as of yet, mainly due to time constraint. Nothing interesting going thru my sig, except for a CECHJ04 (which I forgot to edit the list to show up in there) that has Chemicon stock caps instead of NEC, that I suppose I can try to push a lil' bit.
In the end, I'm calling this a night, it's 3:16 AM here already. I might come back tomorrow if I have any luck with the J04, it needs its fill of games anyways.
TheFellowV
Forum Noob
2504B here running Evilnat 4.91.2 PEX with 600/750 OC, wanted to chime in, see if anyone has similar experience
Gran Turismo 5 Prologue (both release version and when updated to spec III) apparently doesn't like this OC because it hangs, most often at loading screen or when it just fades out of it. When it hangs, it will just cycle parts of frame it last managed to render and output every 2-3 seconds (e.g fps overlay then loading screen leaderboard and back or similar combo). Curiously it looks like games goes on because at one point when I've waited a bit I could hear the game loaded, just that output was shot.
I've played Crysis HD for around an hour (played from start of campaign until you find frozen boat) causing explosions, getting into firefights, generally trying to put stress on the PS3 and I didn't notice any sings of instability or artifacts, neither I could notice anything in Killzone HD/2, Wipeout HD and few others so it looks like its a GT5P problem?
Gran Turismo 5 Prologue (both release version and when updated to spec III) apparently doesn't like this OC because it hangs, most often at loading screen or when it just fades out of it. When it hangs, it will just cycle parts of frame it last managed to render and output every 2-3 seconds (e.g fps overlay then loading screen leaderboard and back or similar combo). Curiously it looks like games goes on because at one point when I've waited a bit I could hear the game loaded, just that output was shot.
I've played Crysis HD for around an hour (played from start of campaign until you find frozen boat) causing explosions, getting into firefights, generally trying to put stress on the PS3 and I didn't notice any sings of instability or artifacts, neither I could notice anything in Killzone HD/2, Wipeout HD and few others so it looks like its a GT5P problem?
Tanzu15
Member
If you install an official firmware, the jailbreak goes away and you have to exploit again if you want to rejailbreak. But yes it's fully reversible.
havenoenterkey2
Forum Noob
Not sure if it's helpful, but I was getting green flash on battlefield 3 when in an online match exiting customization menu for the character. It was only happening on overclock (600/750) and my console (cech2100/cxd5300a1gb) has like 60 days of usage (I bought it in April like new)
Mitsu™
Senior Member
You can switch between OC and non-OC CFW at will. A regular .pup installation as usual and you're good.
I globally agree with you on this (surprising !
).PS3s are fragile consoles. Even Slims and Super Slims.
They were clearly not designed to last for decades.
Reliability was not Sony's priority at all. Just look at the SYSCON curve that makes the fan silence pass before effective cooling.
All PS3 models have weaknesses, even the 25XX. After these, Sony went into "money saving mode". Heatsink increasingly mediocre, smaller fan, etc...
All that to say, it's clear that a PS3 will never have the life expectancy of a NES for example.
So basically, "memento mori". Remember that your PS3 will die
Being aware of this, you have to enjoy it as much as possible while they still work.
Now as you say, good maintenance and cooling are crucial to optimize their lifespan but not overclocking excessively is also something to consider.
With or without overclock, a PS3 is far from eternal, but that doesn't mean you have to go suicidal.
800MHz core is already huge! It's a 60% OC! 950MHz VRAM = ~ 46%.
In general we are more on overclocks of 10 to 20%.
I feel like when we just want to warn people about the long-term risks of high frequencies, you kind of take it as a personal attack.
Again, everyone does what they want, I just find it normal to warn others that this RSX overclock is not magic, there are downsides.
Considering that there are diminishing returns (no matter how hard you deny it, it's factual)...
Then, it's up to them to balance the gains/risks.
Where I agree with you again is on RPCS3. I will always prefer a console to an emulator (no matter how sophisticated it is)
I find it way more pleasant, it's like diving into another era.
No need for 4k/60 FPS to enjoy the classics of this console (I do not spit on a more stable frame rate that said. I wouldn't use that overclock otherwise.).
Last edited:
Raidmax
Member
Just a heads up, I know someone's slim that got to 1680 or so days. While I do admit that one was stock, it was fairly surprising to see a slim run THAT much.
Here it is.
Anyways, back to the thread. 750/975 on a CECHJ04 resulted in some epic fail lol, it was freezing before even showing the waves startup. Back to stock speeds CFW 4.91 (and did I mention navigating the menus in Japanese is fun?) for the time being.
EDIT: Forgot it wasn't a syscon log but a webMAN capture. Edited my text accordingly.
Last edited:
Mitsu™
Senior Member
Only a few PS3 will be able to be stable at 950MHz core and your PS3 isn't. Forget about this frequency.
Even with the best thermal paste possible, it won't allow your PS3 to magically support unstable frequencies.
It's freezing due to the temps, indicating that you're pushing your console to its limits.
It's 100% pointless to keep it that high, unless you manage to keep it cool as ice (water cooling, etc...)
But even then, is it worth it... ?
Last edited:
Edit: You might wanna reinstall the game or delete save files, it happens sometimes.
But as I pointed out in my first ever post here, Gran Turismo actually hammers the GPU a lot more as they have to put out more pixels at 60fps in a forward renderer. Not to mention heavy scenes coming out of loading is like the GPU equivalent of turning on the fire hose.
GT5P and GT5 are 1280x1080 anamorphic with 2xQAA+Temporal.
GTHD is 1440x1080 anamorphic with 2xQAA I believe.
GT6 is 1440x1080 anamorphic with MLAA.
All also feature "overdraw abuse" effects they used to do in PS2 era - things like thick smoke and glare blooms on every light source (yes those are just tons of transparencies).
Crysis HD is 1024x720 anamorphic with many effects buffered sub-native in a deferred renderer typical in a modern engine. In most cases you are bottlenecked by certain parts of the GPU arch and SPU/bandwidth than actually stressing the whole GPU.
Last edited:
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