I was sure you liked to see it again here.
So I would like to see one in my hands.
Thank for the curiosity about view number on you page, I like this kind of geek story.
But RevB of the chip never run on a proto board like that?
yes, but the board has been optimized for chip diagnostics, which means that the communication between chip and host system works for testing chip features.
sort of debug mode and performance is nowhere near production boards.
I did not see Dave's Glaze3D engineering sample while it was for sale, but one thing is still a bit of mystery... main sources claim that original Glaze3D with 8MB eDRAM never in fact materialiazed. Other sources again claim that very very few chips was made, but as it was so much delayed due development of own synthesis tools, they decided not to use time to making it as final product as it would been failure. Instead they started project called AXE which materialized and which I have. It's Glaze3D with stereoids. As from the trademarks filled by the company, AXE was planned to be marketed as Avalanche3D and it supported Dual chip configuration (trademark Avalanche DUAL existed), instead of quad chip of original Glaze. Again, AXE had upgraded THOR (development name for geometry processor which was planned for quad chip Glaze3D.) inside and eDRAM amount was raised to 12MB.
This all went down the drain when Infineon did huge loses in 2001 and decided to kill eDRAM lines and as the AXE was in single chip configuration just an inch faster than GF3, there was no possibility to move the chip to other foundry without losing too much time to stay still competitive. Still Infineon did one batch of B-rev chips to fullfill contract with BB.
The last project which 'boys started for PC was codenamed as Hammer. It was supposed to have some fancy occlusion culling technology (which someone licensed in the end... customer didn't like to see the tech showed in the BB pages so it vanished quite soon after release.) and only 4MB of eDRAM. NEC was the new foundry, but as the guys ran out of money, Hammer never went so far that chips could have been made.
My brain might farting here, but I do recall noticing that ATI's R300 Supertiler system was awfully a lot same way working as Bitboys' eDRAM <--> SDRAM renderer and quite few ppl have speculating if there was some patent issues which gave the boys an edge when negoating the deal... But it's still just speculation as nothing bulletproof has come to proof it.
In any case, IF Dave really did sell original Glaze3D engineering sample, then it would be extremely rare item, as there's no 100% clear confirmation, if such thing even existed. again, if it was an AXE, then the rareness factor comes down a bit, but it's still quite higher than any of main competitors' unreleased hardware.
EDIT: topic says it all, it was an AXE. Congrats to bid winner, it's still pretty rare chip.
Last edited:
