GCC 10 Compiler Drops IBM Cell Broadband Engine SPU Support

[phoronix]

Phoronix: GCC 10 Compiler Drops IBM Cell Broadband Engine SPU Support

Next year's GNU Compiler Collection 10 (GCC 10) compiler release is doing away with support for IBM's Cell Broadband Engine SPU support...

http://www.phoronix.com/scan.php?page=news_item&px=GCC-10-Drops-Cell-BE-SPU

[LoveRPi]

Insanely high performance architecture that developers were too lazy to design for. IBM did everything right except make the barriers to entry low enough for average devs but cache coherency needs to die.

[tildearrow]

Goodbye PlayStation 3

[Alex/AT]

Insanely "high" performance POWER-based architecture with insane development and optimization costs plus no good optimization compatibility with existing software (SPEs are 'units in themselves' that are just pieces of compute power like GPU units that cannot run generalized software). The end result (death of the platform) was pretty predictable.

Also, the general mainstream software compiler optimization for PUs without branch prediction and active microcode execution thread management is a pain, and all such platforms are pretty much doomed.

[CaptainLugnuts]

Insanely high performance architecture that developers were too lazy to design for. IBM did everything right except make the barriers to entry low enough for average devs but cache coherency needs to die.

The SPUs were crap. They were designed by a team that had no idea on how they were to be used and what features were essential.

I spent years writing SPU code for a former game studio. Their only positive was they had control over DMA from the SPU rather than relying on the main CPU to run it. Other than that the EE they supplanted were far superior in design. Wall clock time a single SPU wasn't significantly faster than an EE even though the SPU ran at 8x the clock rate. The SPUs had no swizzling, no broadcast math, and no write masking. You spent all your time screwing around unpacking and repacking the data whereas the EE could use the data without manipulation. The SPUs also had no VIF meaning you couldn't compress vectors in memory and unpack them for free on DMA transfer.

Sony really botched the PS3. It was full of half-assed solutions and buggy hardware.
I did earn really good money dealing with it though, so I can't complain too much.

[s_j_newbury]

I don't really understand why GCC would drop support for a relatively modern, powerful (even if flawed) ubiquitous hardware platform that's still used by enthusiasts just because the vendor no longer supports it. Since when was GCC a vendor oriented compiler project? Is it just a lack of maintainers for the code in question?

[finalzone]

I don't really understand why GCC would drop support for a relatively modern, powerful (even if flawed) ubiquitous hardware platform that's still used by enthusiasts just because the vendor no longer supports it. Since when was GCC a vendor oriented compiler project? Is it just a lack of maintainers for the code in question?

From the article

The Cell BE SPU support was removed from GCC for being "obsolete". Technically this doesn't kill off the Cell BE support entirely as there is still the PowerPC Processor Element (PPC) that uses PowerPC instructions, but the SPE/SPU support is no longer maintained by an open-source compiler toolchain.

When someone will step up to maintain the codes, the support will remain.

[-MacNuke-]

Cell was designed in 2001 when GPGPU, OpenCL and CUDA was not a thing. But it came out pretty late and at that time GPGPU just crushed everything. It was easier to program, faster and cheaper. That's why the PS3 got a proper GPU instead of doing some kind of software rendering on the Cell like the PS1 (GTE) and PS2 (VU) are doing.

[Jumbotron]

The current CEO of AMD had a little something to do with the Cell chip.


MIT Technology Review named her (Lisa Su) a "Top Innovator Under 35" in 2001, in part due to her work with Emerging Products.^[20] Also through the division, she represented IBM in a collaboration to create next-generation chips with Sony and Toshiba. Ken Kutaragi charged the collaboration with "improving the performance of game machine processors by a factor of 1,000", and Su's team eventually came up with the idea for a nine-processor chip, afterwards used to power devices such as the Sony PlayStation 3.

[kpedersen]

I only worked on one project involving the PS3 hardware (it was porting a lesser known gaming title). I found it really interesting but didn't get a massive chance to delve too deep into the hardware or the SPU (the cheap port of the game didn't require too much resources . I would love to have a play again (we even still have one of the devkits laying around).

Slightly unrelated (because access to the SPU is restricted) but I don't believe Sony has much to lose by reinstating the 3rd Party OS functionality. I doubt they have funds to spend on programmers to make this patch though.