I'm sorry I should've added "dedicated" to the multitexturing part too, as I knew about the multi-pass needs. And wasn't the PS2's raw pixel pushing power in the 2000's? 16 pipelines x 147 MHz = 2352 pixels. And how many texture passes were typically performed on average? This I'd be very interested to know. PS2 is so funky, but at the same time very interesting. It's a fun system to mess around with as far as speculation goes with "another 100 MHz on the EE" or "64 MB of RAM vs 32".
Any news on PS2 backwards compatibilty
- Thread starter mjdted
- Start date
I'm sorry I should've added "dedicated" to the multitexturing part too, as I knew about the multi-pass needs. And wasn't the PS2's raw pixel pushing power in the 2000's? 16 pipelines x 147 MHz = 2352 pixels. And how many texture passes were typically performed on average? This I'd be very interested to know. PS2 is so funky, but at the same time very interesting. It's a fun system to mess around with as far as speculation goes with "another 100 MHz on the EE" or "64 MB of RAM vs 32".
I don't know about averages - but from what I remember number of operations per pixel in PS2 (and XBox) exclusives was quite a bit higher then 2001-2002 era PC games.Mobius1aic said:
The highest I'm aware of on PS2 was over 40 operations per pixel (including overdraw), the games I personally worked on were in mid 20ies iirc.
My numbers were just illustrative. I don't know the figures off-hand, but recall the 40x 'overdraw' that Fafalada has mentioned.
Mostly it's the bandwidth. 30GB/s of PS2 edram versus 22 GB/s of PS3 GDDR. That's not actually such a problem though. The real issue is the funny, uncnventional techniques employed by GS. Fafalada has gone into detail on this before and I'm sure a search will dig out his points.
I thought the GS eDRAM memory bus was 48 GB/sec?
Panajev2001a
Veteran
In theory it is around the hundreds of GB/s (150 GB/s IIRC), the 48 GB/s figure counts the bandwidth from the 8 KB page buffers to the pixel engines, but those buffers are filled from the e-DRAM macros much faster than that
.Still the problem of filling page buffers at that high speed should be the easily avoidable one on a platform in which you have almost 8x the GPU available memory as PS2 has total memory (main RAM + VRAM + SPU2 RAM + I/O CPU RAM)... 256 MB vs 40 MB... that is, you can keep all the memory the GS would try to load and not have the problem of these page buffers having to be refilled IMHO.
Extreme mode especulation here...forgive me(if was post before).
If not my mystake ...wee see a patent for graphs useing 8 SPUs.
Is there any possibility for coming ps3 slim with Cell with 8 SPUs(today with better yields) for somekind "full software emulation support" ps2?
If not my mystake ...wee see a patent for graphs useing 8 SPUs.
Is there any possibility for coming ps3 slim with Cell with 8 SPUs(today with better yields) for somekind "full software emulation support" ps2?
Last edited by a moderator:
I thought the eighth SPU was unlocked through firmware once it was figured out the yields were fine, much like a firmware and devkit update allowed PSP devs to use the full 333 MHz capability of the PSPs main processor?
How do you unlock the 8th SPE in earlier PS3's when it's dead? You cannot gaurantee which PS3 has 8 available SPEs, unless a particular SKU was knowingly fitted with 8 SPE Cells and can be used as such.
I wasn't aware they actually "killed" it, I thought it was just left unused in case one of the SPE was to die from a manufacturing defect, in which case the 8th SPE would be left as a back up but still usable kinda like Phenom II x3 CPUs with the full 4th core still intact in many of them.
They probably didn't kill it, but wasn't Cells with 8 working SPE used for IBM boards? And 7 working SPE used for PS3? Ie the chance of the 8th SPE actually not being defective is slim?
Also the patent shows use of 8 SPE, but does it need 8, can it run on 6 or 7?
Because the 1 SPE is reserved for security and 1 is disabled,/broken ie there is only 6 guaranteed available for use in the PS3.
The 60GB with full hardware backwards compatibility would have to have been the only model that used a 7 SPU Cell.
While I don't think it's true, I don't follow your logic.
I didn't say they killed it, only that it was dead! Among the PS3 population, there are PS3's with 7 viable SPEs and PS3's with 8 viable SPEs. To keep everything equal, those with 8 viable SPEs have one deactivated. If SPE8 was to be reactivated, you'd end up with some PS3's more powerful. If used for BC only, you couldn't be sure which PS3's would gain the advantage of SPE8. Unless 8 SPE capable PS3's are a particular SKU. eg. Every 40 GB PS3 could in theory be equipped with an 8 working SPE Cell, with one SPE deactivated at the moment but which could be unlocked for BC. That could work.
DuckThor Evil
Legend
I think the whole idea where somehow the last spu is the key and critical part to 100% software emulation of PS2 on PS3 is just flat out retarded. I mean speculating things is fine, but come on.
I understand the whole old 7 SPE Cell vs. 8 SPE Cell issue, I'm just trying to get down to whether or not using the 8th SPE in the first Cells was possible. How is it dead? Just a manufacturing defect in the lithography with expectations of 1 SPE not being formed right that makes it un-usable, leading Sony to just use Cells using 7 SPEs since they possibly expect one to not work?