SGI: Hardware

hi
I got an email from a dude who claims he has installed a - R12K - R-twelve-key - CPU - on his Indigo2 (isn't limited to R10K @ 195 Mhz?), so … I wonder if it's a bullshit of if it is still existing such a kit/hack/whatever it is

never heard, so guys, experts, I give you the microphone

http://lmgtfy.com/?q=indigo2+r12000+site%3Anekochan.net

first hit

WTF? I wonder WHY? my searching things is always not working on this damn Safari-MacOSX/Lion box!

before opening this thread I tried with google-dot-it without any result
and also, this thread was not listed by the forum search function
while you have got it as the first hit! damn, it makes me a bit frustrated!

btw, thanks for the link

It must be your Italian googletzi..

uunix wrote: It must be your Italian googletzi..

ummm it may be, I am experimenting google dot com instead of google dot it :
- searching for " Avoget/AvCase " is producing results instead of nothing (and what I get it's not porn )
- and I got less frustration with broken links

dot com >> better >> dot it

Indigo2 cpu_module hack: can we predict the expected speedup between an R10K@195Mhz and an R12K@200Mhz ?

the difference is just the ISA, new instructions and more efficiency in the R12K core
I do not expect a great difference

ivelegacy wrote: Indigo2 cpu_module hack: can we predict the expected speedup between an R10K@195Mhz and an R12K@200Mhz ?

the difference is just the ISA, new instructions and more efficiency in the R12K core
I do not expect a great difference

R12K had a few enhancements to branch prediction that likely helped a bit (e.g., global history), but the real kicker would be the Delay Speculative Dirty (DSD) bit added to the CP0 config register, which would be useful in helping to offset some of the side effects of speculative execution, which affects non-coherent machines like IP28 (and IP32). Would still need to back that up w/ cache barriers and, I think, some bounce buffers, but it'd make running Linux on that platform less of an annoyance.

thanks, R1xK is still out of my knowledge, I am playing assembly with R2k/3K and mips32

I bought an Atlas board (1), it comes with a mips32r2 core BE/LE + fpga and LA (logic analyzer) test points, it also inludes (on the fpga side) a cache engine and a burst engine (synchronous data transfer), plus a gdb stub in ROM and a full E-jtag , this one mounted on a common 2.54" pins header. So It's the best platform to experiment the mips architecture because everything is under control.

I do not have an indigo2 , one of my friends has one, but he is not interested in any linux and cpu hacks, he is fine just with Irix and his Illustrator , So I am collecting more informations(2) about the R12K on indigo2 hack , just to be more confident it is really possible, and it really does make sense. I am not able to say a word about the speedup from R10k to R12K , and I dot have the possibility to check by myself


(1) MIPS Technologies , MIPS32 Atlas SDB CPCI Development Board
(2) I am waiting from a reply from the author of the hack,already contacted

Still waiting for a few snapshots from the mod author

as far as I have understood mr tooR has modded a Silicon Graphics Indigo2 CPU Module R10K @ 195Mhz SGI P/N 030-0966-004

• removing the R10K
• replacing with an R12K extracted from an O2+
• desoldering the crystal @ 48Mhz, replacing with a 50Mhz (4x50Mhz=200Mhz)
• it's not clear how he changed the Vcore voltage

does someone has a detailed photo of an original R10K CPU module ?
I'd like to see if I can recognize a switching + LDO stage somewhere