RDRAM chips
-
- Posts: 4
- Joined: Thu Jan 29, 2004 3:26 pm
RDRAM chips
Let me preface by saying I don't even have a PS2 yet, and am therefore practically useless. That being said, i'm getting one shortly(it's in the mail) along with the linux kit and I might be making marcus' serial cable. My question regards the main system RAM(32mb). As far as I've read these are two RDRAM chips soldered directly to the mainboard, I think they're the two to the outside of the EE when looking at the bottom of the board(from DMS3 pics). Questions:
1. What clock are these chips run at?
2. Do they use the standard 16-bit RAMBUS channel?
3. Are they available in 32mb, if each is currently 16mb(*2 = 32mb)?
4. Am I being idiotic?
5. Extending the traces to include two more 16mb chips would bring total up to 64mb, but would increase latency. Would this outweigh the benefit of having 64mb on board?
6. Is there something in the DMAC that limits the RAM to 32mb?
7. where can I find answers to these questions on the web if they're available, and/or you guys don't want to answer them.
oh, and 8. who came up with [email protected]? that's pretty wild.
-Thanks in advance, I know these are strange, possibly retarded questions, but i'm eager to mess with this thing.
-unkle
1. What clock are these chips run at?
2. Do they use the standard 16-bit RAMBUS channel?
3. Are they available in 32mb, if each is currently 16mb(*2 = 32mb)?
4. Am I being idiotic?
5. Extending the traces to include two more 16mb chips would bring total up to 64mb, but would increase latency. Would this outweigh the benefit of having 64mb on board?
6. Is there something in the DMAC that limits the RAM to 32mb?
7. where can I find answers to these questions on the web if they're available, and/or you guys don't want to answer them.
oh, and 8. who came up with [email protected]? that's pretty wild.
-Thanks in advance, I know these are strange, possibly retarded questions, but i'm eager to mess with this thing.
-unkle
We had access to too much equipment, too much money, and little by little, we went insane.
1. No idea, I'm sure someone knows though. My guess would be at the same clock as the EE's system bus.
2. Who knows? Are you able find out anything about them based on what's printed on the chips?
3. Google.
4. I'm curious as to how you plan to desolder those chips (iirc they are BGA) or "extend the traces".
5. No idea :) (seems a repeating theme).
6. No. Physical RAM is 28-bits (or is 26?), that's as much as you can address in a 32-bit address space.
7. Google.
2. Who knows? Are you able find out anything about them based on what's printed on the chips?
3. Google.
4. I'm curious as to how you plan to desolder those chips (iirc they are BGA) or "extend the traces".
5. No idea :) (seems a repeating theme).
6. No. Physical RAM is 28-bits (or is 26?), that's as much as you can address in a 32-bit address space.
7. Google.
The RDRAM chips run at 300 Mhz (AFAIK) just like the EE.
The VUs run at 300Mhz too. GS, VIF and GIF run at halt ie 150Mhz.
The official Sony dev kit has 128Mb RAM so yes you can theoretically have more, BUT, there is no room on the retail motherboard.
My v9 PS2 seems to have only 1 RDRAM chip.
It's a dual channel architecture.
It's the cheapest and crapiest RDRAM btw. They spec'ed it back when 800mhz RDRAM was cutting edge still.
You can probably lots of answers by reading the Sony manuals that come with the Linux kit (or if you find them online somewhere).
The VUs run at 300Mhz too. GS, VIF and GIF run at halt ie 150Mhz.
The official Sony dev kit has 128Mb RAM so yes you can theoretically have more, BUT, there is no room on the retail motherboard.
My v9 PS2 seems to have only 1 RDRAM chip.
It's a dual channel architecture.
It's the cheapest and crapiest RDRAM btw. They spec'ed it back when 800mhz RDRAM was cutting edge still.
You can probably lots of answers by reading the Sony manuals that come with the Linux kit (or if you find them online somewhere).
-
- Posts: 4
- Joined: Thu Jan 29, 2004 3:26 pm
Ok, i did some research, and put the ram idea on the back burner for now. Probably revisit it when I can get some grant funding. If you're interested, a couple of people have been able to develop techniques for soldering BGAs in a home/research environment(i.e. low-budget). Here's the links:
University of Melbourne made their own PowerPC dev boards for the robo soccer cup: BGA mounting at the bottom of the page:
http://www.cs.mu.oz.au/tsa/projs/ppc/
And a german guy wrote a page on mounting BGAs (in german) in a one-off environment:
http://wwwbode.cs.tum.edu/~acher/bga/index.html
SO... it is possible, and if i really, really felt so inclined, i'd get on it, but as it is, i'll accept the challenge of using 32mb. Next project: writing a driver to use marcus' serial cable to directly connect two ps2 mobos. I want to run openmosix on the pair, but ethernet doesn't strike my fancy. I figure if the connects are under 4" long, oxygen-free copper cabling, I should be able to get decent interconnect speeds. you are all free to write me off for now, but I hope I can get back in 6mo with a dual mobo ps2 workstation.
-peace
University of Melbourne made their own PowerPC dev boards for the robo soccer cup: BGA mounting at the bottom of the page:
http://www.cs.mu.oz.au/tsa/projs/ppc/
And a german guy wrote a page on mounting BGAs (in german) in a one-off environment:
http://wwwbode.cs.tum.edu/~acher/bga/index.html
SO... it is possible, and if i really, really felt so inclined, i'd get on it, but as it is, i'll accept the challenge of using 32mb. Next project: writing a driver to use marcus' serial cable to directly connect two ps2 mobos. I want to run openmosix on the pair, but ethernet doesn't strike my fancy. I figure if the connects are under 4" long, oxygen-free copper cabling, I should be able to get decent interconnect speeds. you are all free to write me off for now, but I hope I can get back in 6mo with a dual mobo ps2 workstation.
-peace
We had access to too much equipment, too much money, and little by little, we went insane.
You're not wrong, but you're heading in the wrong direction...
Firewire (1394) runs at 400Mbps. USB (1.0) runs at 11Mbps. Ethernet runs at 100Mbps. Nice, fun speeds that are unfortunately limited by the speed of the processor that talks to them - the IOP. Good luck reaching anything close to that. (BTW, those are maximum speeds, theoretical in nature and rarely reached in the best of cases...)
Serial runs direct from the EE. There's no fooling with SIF access and the slow, slow IOP. The problem is it is only serial. Very slow. But at least its easier to use. :)
Firewire (1394) runs at 400Mbps. USB (1.0) runs at 11Mbps. Ethernet runs at 100Mbps. Nice, fun speeds that are unfortunately limited by the speed of the processor that talks to them - the IOP. Good luck reaching anything close to that. (BTW, those are maximum speeds, theoretical in nature and rarely reached in the best of cases...)
Serial runs direct from the EE. There's no fooling with SIF access and the slow, slow IOP. The problem is it is only serial. Very slow. But at least its easier to use. :)
I can sense a lot of confusion starting in this thread :-P Kbps which magically becomes Mbps, serial io which magically becomes IEEE 1394, and all of that thanks to Cloud ;-)
pixel: A mischievous magical spirit associated with screen displays. The computer industry has frequently borrowed from mythology. Witness the sprites in computer graphics, the demons in artificial intelligence and the trolls in the marketing department.