STM Kyro3: was it never taped out?
- Thread starter Simon82
- Start date
Apparently, yes, as a colleague just waved a development board at me
Panajev2001a
Veteran
Apparently, yes, as a colleague just waved a development board at me
They must love you at the office, were you able to dodge it in time ?
.Megadrive1988
Veteran
Kyro, Kyro II were of the same generation, PowerVR Series3 IIRC
Kyro III was then, the next-gen, Series4 with a completely new architecture including on chip T&L / geometry engine
(or Vertex Shader), if I'm not mistaken.
Many of us were hoping for a PowerVR Series4-based SEGA Dreamcast2 and NAOMI 3 arcade board based on it.
Kyro III was then, the next-gen, Series4 with a completely new architecture including on chip T&L / geometry engine
(or Vertex Shader), if I'm not mistaken.
Many of us were hoping for a PowerVR Series4-based SEGA Dreamcast2 and NAOMI 3 arcade board based on it.
I think it had a T&L unit and 4 pixel pipelines.
Well, it's been quite a while. It's not like we expect old men to remember such things...
Perhaps you'll also do us the favor of also failing to fully remember why it never came to be as a product? Because STMicro backed out of graphics? Or did STMicro back out because PowerVR 4th gen. would not have made it to market? Something else?
Perhaps you'll also do us the favor of also failing to fully remember why it never came to be as a product? Because STMicro backed out of graphics? Or did STMicro back out because PowerVR 4th gen. would not have made it to market? Something else?
Ha! A few months of sleep deprivation is the real killer.Well, it's been quite a while. It's not like we expect old men to remember such things...
(Note: Not an official IMG response) That's my understanding. It's a real shame <shrug>.
Looking at MBX's VGP I have the suspision it wasn't "just" a T&L unit; you think hm? Bares the question who actually designed the algorithms for those things LOL
That may work for your subordinates ....
I don't recall the details but I think it was a pre-DX8-style T&L unit. That, of course, has both advantages and disadvantages.
Panajev2001a
Veteran
How much of it was related to the R&D work that went into Elan ?
Megadrive1988
Veteran
I wish that it had been PowerVR Series3 (KYRO, KYROII) that had been the one architecture that got canceled, instead of Series4. I wish that Series4 had been released instead, and powered upper midrange to highend PC cards.
Man, it seems like all the good, promising stuff gets canceled, or never makes it to the consumer/gamer market
*Lockheed Martin: Real3D/100 (professional only)
*Rendition: 'Thriller Conspiracy' (Fujitsu FXG-1 "Pinolite" geometry processor + V2200),
Vérité V4400E
*3DO Systems / CagEnt: Bulldog/M2 (arcade + other units but not console or PC card), MX/M3 (almost redesigned for "Nintendo 128"),
S42/M4 (?)
*TriTech / BitBoys: Pyramid3D TR25201
*BitBoys: Glaze3D, Axe, Avalanche3D, Hammer
*3Dfx: Rampage / Spectre / Sage / Tantrum / Fear / Sage2 / Fusion / Fearless / Mojo
*Videologic / ImgTec: PowerVR Series4
Man, it seems like all the good, promising stuff gets canceled, or never makes it to the consumer/gamer market
*Lockheed Martin: Real3D/100 (professional only)
*Rendition: 'Thriller Conspiracy' (Fujitsu FXG-1 "Pinolite" geometry processor + V2200),
Vérité V4400E
*3DO Systems / CagEnt: Bulldog/M2 (arcade + other units but not console or PC card), MX/M3 (almost redesigned for "Nintendo 128"),
S42/M4 (?)
*TriTech / BitBoys: Pyramid3D TR25201
*BitBoys: Glaze3D, Axe, Avalanche3D, Hammer
*3Dfx: Rampage / Spectre / Sage / Tantrum / Fear / Sage2 / Fusion / Fearless / Mojo
*Videologic / ImgTec: PowerVR Series4
Last edited by a moderator:
Second that; but I personally wasn't that disappointed over Series4 in the end as I was over the cancellation of the Series5 GPU. While Series4 was to be another mainstream GPU, Series5 was feature-packed enough to get called a high end GPU if they would had released it on time. A mini version of that thingy is of course today SGX, but there are lightyears apart in terms of performance between small devices like handhelds, PDAs or even mobile phones and a PC GPU.
....or maybe there's always a large myth surrounding all the fore mentioned vaporware, which might have never realised to its expected potential in the end. Paperspecs can be highly misleading and I severely doubt that any of those would had been able to perform any wonders. Highly competitive in their own areas most likely yes, but I have severe doubts that any of those would had trounced any of the GPUs across the board that made it on shelves after all.
Series4 might had given the GF4 of the time a good run for its money, but they most certainly would had missed on the few spare pixel shaders that were present in some games at the time. Series5 would had been far more interesting from many aspects; one would be that it wouldn't had been the first high end TBDR available and the close second that we would had seen if anything SM3.0 and beyond bears any advantages to deferred renderers.
Anyway Imagination is a tiny company compared to AMD's GDP or NVIDIA. However most if not all the stuff they develop is outstanding IP. It's not just the stuff from PowerVR that's relatively impressive; Ensigma and Metagence IP are highly interesting too. It's a bloody shame that Intel probably won't ask for stuff with higher potential then for UMPCs.
Well at least they managed to stay alive and not be absorbed up to now and for the first time in many years they turned actually profitable for this year and if they're forecasts are true and they reach their goals for their target markets, their future seems even brighter than today.
more of them:
GigaPixel GP1/2 (free FSAA, faster than anything else when using FSAA 4x), GP3 (EMBM, 8-layer, T&L, DXT1, 80KB SRAM)
OAK WARP5 (Windows accelerator and Rendering Processor) - 1997, trilinear, free FSAA
Stellar PixelSquirt - 1998, scan-line rendering (per-pixel tile-based rendering), 32bit rendering+Z, extremely low tranzistor count, promised performance around TNT level
Philips Trimedia TM1 - VLIW architecture, full-hardware acceleration of MPEG1/2 (Philips acquired graphics division of Western Digital)
Alliance Paladin - the only non-3Dfx Glide compatible accelerator
GigaPixel GP1/2 (free FSAA, faster than anything else when using FSAA 4x), GP3 (EMBM, 8-layer, T&L, DXT1, 80KB SRAM)
OAK WARP5 (Windows accelerator and Rendering Processor) - 1997, trilinear, free FSAA
Stellar PixelSquirt - 1998, scan-line rendering (per-pixel tile-based rendering), 32bit rendering+Z, extremely low tranzistor count, promised performance around TNT level
Philips Trimedia TM1 - VLIW architecture, full-hardware acceleration of MPEG1/2 (Philips acquired graphics division of Western Digital)
Alliance Paladin - the only non-3Dfx Glide compatible accelerator
Megadrive1988
Veteran
Forgot about those, thanks.
I do now remember reading that SEGA had looked at PixelSquirt, as well as Real3D. I assume PixelSquirt would've either been used in the never-released Saturn 3D upgrade, or, in place of the 3Dfx chip in Black Belt, or in place of PowerVR2 in Katana/Dreamcast.
Last edited by a moderator:
the maddman
Regular
I have a Warp5. You only got "free" FSAA with extreme texture compression. Gave everything a funny pink/purple tinge. No OpenGL drivers for it either.
It was pretty slow without it's texture compression, maybe about half a voodoo with it on.
