Is there a way to determine the manufacturer by examining the RSX string?
I think we could ask
@Evilnat if he can add a function to his XAI plugin to display the information.
The FAB ID was spelled out by testbench firmware, which is how we now know the Fabs that founded the silicon die are 1 = TOSHIBA, 2 = SONY and 4 = TSMC. Now that we know this information, sure it could be built into CFW and various tools.
@M4j0r might know more about the other information contained in that tuplet (bracketed string). There is clearly more information contained there.
D/S line I think stands for "design/specification" line, which would be the specific fabrication line it came off of at that FAB manufacturer. TSMC for example has many, since they're the largest semiconductor mnf in the world.
Sony and Toshiba were a joint venture on the RSX and SONY sold TOSHIBA their 1st floor DS line at Nagasaki FAB2. So TOSHIBA chips were made on the 1st floor and SONY chips were made on the 2nd. Same building complex, different owners, but they would have both been packaged by the same place. Where the die was soldered to the interposer and underfill was applied. Most likely at Oita TEC. TSMC has their own facilities and packaging partners.
Could you elaborate on what we need and provide more details so Nat can modify the XAI plugin to display only the necessary information? I was specifically looking for the first code, AP00008286. What does it mean? At least I found that the codes starting with APXXXXXXXX are the ones that handle higher overclocks.
This is our outdated table (i know, i need to update as he got lots of new reports_
Outdated Table
| RSX | Rev | Max GPU Clock (Stable) | Max Memory Clock (Stable) | Full Details | Notes | PS3 Model | Owner |
|---|
| rsx40 | a01 | 850 MHz | 1000 MHz | vpe:ff shd:77 [AP0022017:0:1:3:e:18:2:0:1][2a:0:a:0:1:0:1][1:1:0] | Random crashes at 900 core (Only on RSX-demanding games)
Can go up to 1000/1000 without bricking. | CECH-2511A | Me |
| rsx40 | a01 | 850 MHz | 1000 MHz | vpe:ff shd:3f [AP0029328:0:2:11:16:20:2:0:2][24:0:a:0:1:0:1][1:1:0] | CECH-2511A
Can go up to 900/1000 without bricking.
(Not tested on highter core) | @JoeSapecudo | |
| rsx40 | a01 | 700 MHz | 1000 MHz | vpe:ff shd:3f [NN8941-24:0:4:18:12:18:6:0:2][25:0:a:0:1:0:1][1:1:0] | Artifacts at 750Mhz Core in specific games. | CECH-2504A | @Thimblewad |
| rsx40 | a01 | 650 MHz | 800 MHz | vpe:ff shd:3f [NN0078-06:0:4:6:18:4:6:0:2][25:0:2:0:1:0:1][1:1:0] | Artifacts at 700Mhz Core in specific games.
Black Dots and crashes at 950 Mhz memory. | CECH-2104B | @cha0shacker |
| rsx65 | a06 | 700 MHz | 800 MHz | vpe:ff shd:6f [KBA857800:1:2:c:11:7:3:8:2][3d:0:2:0:1:2:0][2:1:0] | Artifacts at 750Mhz Core in specific games.
800Mhz memory is stable.
WARNING: 1000 Mhz memory bricked his system. | CECH-2004B | @Mitsu™ |
| rsx40 | a01 | 700 MHz | 1000 MHz | vpe:1bf shd:3f [NN9095-14:0:4:e:3:12:6:0:2][1e:0:a:0:1:0:1][1:1:0] | Artifacts at 750Mhz Core in specific games. | Frankenstein CECH-A00 | @XDeltaOne |
I have been keeping track of this info for a while and grabbed any unique RSX info dumps shared, but I haven't read the entire thread yet. It expands as fast as I can read it!
Something I feel compelled to say is I doubt there's a connection between the date code and higher clocking 40nm RSXs. So far we don't have enough lvl1 dumps with the RSX string of people with 25xx models at their highest OC to establish a trend. I can confirm there are 40nm RSXs made by TSMC, SONY, and TOSHIBA. But only a couple reports from people with high clocks.
If people with a BRICK WISH want to help me to see if there is a trend here, I need you to...
- Find you highest stable overclock
- dump lvl1 using CFW tools
- Open the lvl1 dump it in HxD and search for your clock speeds. For example, if you have default clocks search for "500/650" and it'll take you right to this info...
- Post that in this thread with the following information.
- CFW tools --> dimp tools --> dump SYSCON error log. This contains the platform ID which I can use to find your model number and motherboard revision. It also containg the console's uptime, how much it's been used. Newer chips likely can clock higher. The errorlog may register YLOD events related to instability or insufficient filtering. This is useful information.
- MNF Date/Code
- Color of the die if you happen to know it
- Model number of your RSX if you know it. It's written on the IHS of your RSX. If you clean the paste or delid (only if know what UR doing) pay attention to this.
Here's what I have so far....
| RSX | Node (nm) | DIE | FAB | MFG | DS | VIDBin | MODEL No. | MB rev. | lvl1 RSX String | DAYS |
| ? | 40 | ? | 2 | SONY | 3 | 0 | CECH-2500 | ? | rsx40 a01 600/750 vpe:ff shd:3f [L6B882200:0:2:16:10:5:3:0:2][1d:0:a:0:1:0:1][1:1:0] | 70 |
| ? | 40 | ? | 2 | SONY | 3 | 0 | CECH-2501B | JSD-001 | rsx40 a01 900/950 vpe:ff shd:3f [L4B576500:0:2:14:c:17:3:0:1][2f:0:a:0:1:0:1][1:1:0] | 42 |
| ? | 40 | ? | 1 | TOSHIBA | 2 | 0 | CECH-2511A | ? | rsx40 a01 600/750 vpe:ff shd:77 [AP0022017:0:1:3:e:18:2:0:1][2a:0:a:0:1:0:1][1:1:0] | 39 |
| ? | 40 | ? | 1 | TOSHIBA | 2 | 0 | ? | ? | rsx40 a01 600/750 vpe:ff shd:3f [AP0023203:0:1:d:17:1a:2:0:1][26:0:a:0:1:0:1][1:1:0] | NA |
| CXD5301DGB | 40 | Silver | 4 | TSMC | 6 | 0 | FRANKY | COK-002 | rsx40 a01 500/650 vpe:ff shd:3f [N4M735-23:0:4:17:14:c:6:0:1][24:0:e:0:1:0:1][1:1:0] | |
| ? | 40 | ? | 4 | TSMC | 6 | 0 | CECH-2100 | SUR-001 | rsx40 a01 600/750 vpe:ff shd:3f [NN2391-04:0:4:4:c:6:6:0:2][23:0:a:0:1:0:1][1:1:0] | |
| ? | 40 | ? | 4 | TSMC | 6 | 0 | CECH-25xx | JSD-001 | rsx40 a01 600/750 vpe:ff shd:3f [NN9634-19:0:4:13:4:f:6:0:2][27:0:e:0:1:0:1][1:1:0] | |
| CXD2982BGB | 65 | Blue | 2 | SONY | 3 | 0 | CECHK06 | DIA-002 | rsx65 a03 1f4/28a vpe:ff shd:3f [H8D341900:1:2:16:f:c:3:7:1][1c:0:2:0:1:2:0][2:1:0] | |
| ? | 65 | ? | 2 | SONY | 3 | 0 | CECH-Lxx | VER-001 | rsx65 a06 600/750 vpe:ff shd:3f [J5E248000:1:2:a:c:9:3:8:2][1d:0:2:0:1:2:0][2:1:0] | |
| ? | 65 | ? | 2 | SONY | 3 | 0 | CECH-Lxx | VER-001 | rsx65 a06 700/800 vpe:ff shd:3f [J2D948400:1:2:8:d:b:3:8:1][1c:0:2:0:1:2:0][2:1:0] | |
| CXD2982GB | 65 | Pink | 1 | TOSHIBA | 2 | 0 | CECHJ02 | DIA-002 | rsx65 a05 1f4/28a vpe:ff shd:3f [AP0009984:1:1:e:13:a:2:5:1][1c:0:2:0:1:2:0][2:1:0] | |
| CXD2982GB | 65 | Blue | 1 | TOSHIBA | 2 | 0 | CECHK02 | DIA-002 | rsx65 a05 1f4/28a vpe:ff shd:3f [AP0009644:1:1:15:10:6:2:5:2][3d:0:2:0:1:2:0][2:1:0] | 174 |
| CXD2991GB | 65 | GOLD | 1 | TOSHIBA | 2 | 0 | FRAMED | VER-001 | rsx65 a05 1f4/28a vpe:ff shd:6f [AP0012532:1:1:15:14:a:2:5:2][1c:0:2:0:1:2:0][2:1:0] | NA |
| CXD2991BGB | 65 | Blue | 1 | TOSHIBA | 2 | 0 | CECH-K02 | DIA-002 | rsx65 a05 1f4/28a vpe:ff shd:3f [AP0010333:1:1:9:15:14:2:5:1][1c:0:2:0:1:2:0][2:1:0] | 173 |
| CXD2991GB | 65 | ? | 1 | TOSHIBA | 2 | 0 | CECH-P01 | VER-001 | rsx65 a06 650/800 vpe:ff shd:77 [AP0011387:1:1:c:10:b:2:5:2][1d:0:2:0:1:2:0][2:1:0] | |
| ? | 65 | ? | 1 | TOSHIBA | 2 | 0 | CECH-Kxx | DIA-002 | rsx65 a05 600/750 vpe:ff shd:3f [AP0009981:1:1:18:a:c:2:5:1][1c:0:2:0:1:2:0][2:1:0] | |
| CXD2971-1GB | 90 | Blue | 2 | SONY | 3 | 3 | CECHG02 | SEM-001 | b07 1f4/28a vpe:ff shd:7e [H3C662200:1:2:a:f:b:3:12:2][16:3:2:0:1:3:1][0:0:0] | |
| ? | 90 | ? | 2 | SONY | 3 | 2 | DECR-1000A | TMU-520 | b03 600/775 vpe:ff shd:3f [G8B644200:1:2:13:8:a:3:f:1][39:2:0:0:1:3:1][0:0:0] | |
| CXD2971DGB | 90 | Blue | 1 | TOSHIBA | 2 | 3 | CECHG02 | SEM-001 | b08 500/650 vpe:ff shd:3f [AP0008495:1:1:b:10:c:2:11:1][16:3:1:1:1:3:1][0:0:0] | |
| CXD2971DGB | 90 | Blue | 1 | TOSHIBA | 2 | 3 | CECHC02 | COK-002 | b08 500/650 vpe:ff shd:3f [AP0006466:1:1:13:b:12:2:f:2][16:3:1:1:1:3:1][0:0:0] | |
Side note:
my CECH-2501a has a CXD5300 40nm RSX. That's the 1st revision of 40nm RSX there are. The 30xx model tends to have a CXD5301 revision and early SS models have the CXD5302. Since those later revisions are in consoles that cannot be overclocked without full custom firmware the only way to see if they can overclock higher is to remove them and solder them onto an earlier model. I do not want to condone the harvesting of 40nm RSX to put onto other 40nm containing models. So what I'm talking about here is the Frankenstein Phat mod.
If there are efficiency improvments in these later model revision 40nm RSXs, a backwards compatible model with a new old stock CXD5301 for example should be able to clock higher than any 25xx model. It has a larger heatsink meant for a 90nm after all. And it has a voltage regulation designed for more power hungry chips. At high clocks a frankie should be able to deliver more stable voltage and dissipate more heat.
...besides, I think I am the only person on earth, who does explain technical stuff in layman-terms, as easy as possible, so even a young child can understand me
Hey! Take it back! ;p