someone (someone@somewhere.com) on 9/8/09 wrote:
---------------------------
>AM (@.) on 9/7/09 wrote:
>---------------------------
>>Jouni Osmala (josmala@cc.hut.fi) on 9/6/09 wrote:
>>---------------------------
>>>>>>Power and space yes, the question is by how much. E.g. PPC750 and esp. 603 consumed
>>>>>>only a few watts and had a very modest non-cache transistor count by any standards.
>>>>>>What do you think power and pipeline-length overheads are there due to ooo?
>>>>>
>>>>>And both clocked for lower than other chips made in the same process tech...
>>>>
>>>>This is not true at all, and in case you have forgotten something Jouni, please
>>>>refresh your memories from references. E.g. here is some info from Motorola PowerPC
>>>>603e microprocessor fact sheet (you should find one easily):
>>>>
>>>>200 MHz @ 0.35 um, 81 sq.mm, 4/5 W typ/max
>>>>300 MHz @ 0.29 um, 42 sq.mm, 4/6 W typ/max
>>>
>>>600 MHz @ 0.35um EV56
>>>And intel got 300MHz @ 0.35 (klamath)
>>>
>>>So PPC WAS lower clocked, due to fact it was short pipeline.
>>>So it isn't really a proof that with OoO you can have both short pipeline AND clock it high.
>>
>>Logic reasoning and facts are hardly friends of yours, ain't they? You mention
>>Alpha; it clocked at ~600 MHz in 0.35 um for both '164 and '264 (IIRC 575 MHz),
>>same pipeline length, and it was way above *anyone else* in the industry, in-order or OoO, wasn't it?
>
>EV56 (0.35 um) @600 MHz shipped in volume and had a 45 W TDP
>
>EV6 (0.35 um) @500 MHz shipped in volume and had a 91 W TDP
>
>There was a 575 MHz EV6 disclosed as an expensive option
>for the biggest Turbolaser system DEC sold and that SKU
>had a 107.5 W TDP. In no way can 575 MHz be considered a
>volume SKU for EV6. I have my doubts whether it actually ever
>sold commercially. IMO it is in the same category as the 767
>GHz chilled EV56 system that was offered briefly by a third
>party vendor before EV6 became generally available.

Well, I think volume doubts can be expressed about any high-end representative of any arch, be it Alpha, Itanium, Power etc.; I really don't see what makes EV6 any more subjectable to it than others.

Interestingly, quick search shows that even the very first offerings of DEC's servers with EV6 were 575 MHz, and unlike many other high-end offerings, their pricing was made public: http://news.cnet.com/Compaq-rolls-out-Alpha-server/2100-1001_3-216826.html Sure, $200k ain't cheap, but as I have just found out, as early as June'99 one could buy an EV6 box for as little as $3500 (@466 MHz)! http://www.thefreelibrary.com/New+Compaq+AlphaServer+DS10+is+World%27s+Fastest+Single+Processor+RISC...-a054589432 So, if true, I really don't understand when some people say how '264 boxes were completely unaffordable for the masses. I would have bought one w/o thinking to run simulations, I wonder if it had any competition back then at such price/performance.

And unless I missed some info, fastest EV56 that DEC servers ever sold with appears to be 612 MHz, so how do you conclude that the volume argument applies to one fastest available chip, but not another? After all, early EV6 pubs (Gwennap's Oct'96 column, Leibholz & Razdan's compcon presentation, Keller's own slides from '96 or early '97) describe it as a 500 (500+) MHz device -- an indication that if anything, the team overdelivered on the promises.

The 107.5 W figure you mention -- my copy of '264 HRM also includes 600-MHz @ 109 W, and refers to it as device's maximum or peak power. Other Alpha tech papers characterize original EV6 as 7x W @ 600 MHz device, so the 100+ W figures in hrm are probably worst-case figures (worst silicon + most power-drawing code) that MB designers must ensure in their boards, no?