News - AMD apparently disinterested in CPUs for workstations

[HEXUS]

Company remains competitive in professional graphics, but has it given up in processor market?

Read more.

[aidanjt]

Workstation - Server
Xeon - Xeon
Opteron - Opteron

What did I miss?

About the only difference between workstation and server machines is the motherboard, and even those differences range from minimal, to non-existent. One could argue that the 'workstation' motherboard selection for the AMD platform is poor, and I could agree to that. But the professional CPUs are the exact same thing for both platforms whether the machine roll is server or workstation.

Gotta love market 'analysts' who don't even understand their own market.

[kingpotnoodle]

Workstation - Server
Xeon - Xeon
Opteron - Opteron

What did I miss?

About the only difference between workstation and server machines is the motherboard, and even those differences range from minimal, to non-existent. One could argue that the 'workstation' motherboard selection for the AMD platform is poor, and I could agree to that. But the professional CPUs are the exact same thing for both platforms whether the machine roll is server or workstation.

Gotta love market 'analysts' who don't even understand their own market.

I think what they're saying (and here is the usual Peddie bloody obvious) is that as as most manufacturers have stopped selling AMD based workstations (no doubt because they were slow and didn't sell) that AMD have sensibly not put large marketing budgets onto the case and have thus been deemed to have "given up" on the segment. Logical, better to spend your money getting a % of a higher volume market where you stand a chance...

[Alex Herrera]

aidanjt, the workstation platform and markets are very different than what you may remember from years past. There is a segment of the market (dual-socket midrange and high-end) in which the workstation platform does share common components with dual-socket servers (but sorry, the I/O requirements still require a different chipset, different selection of slots and a host of other mods, like power supplies). But the most important thing to emphasize is that that segment represents only about 20% of the market, a limitation which is one of the primary reasons AMD's success in the market was handicapped from day one (due to more of a choice from AMD, not any particular lacking technology or capability).

The bulk of the market today is in mobiles and entry-level workstations, which have a lot in common with high-end notebooks and high-end desktops, not servers. In particular, the biggest growth opportunity going forward is in "low-entry" workstations (e.g. HP Z200 and Dell T1500) which are minimizing costs in an attempt to lure new cost-sensitive user communities (think AutoCAD) into the workstation market.

And as far as AMD being so wise to address other markets they're better equipped to win, that's very arguable. What may not be so "bloody obvious" kingpotnoodle is that entry workstations - at around 60%, the biggest chunk of volume - and especially low-entry workstation (where the growth is) may be one of the markets best suited to AMD. Moving to Fusion and Sandy Bridge generations, AMD has one significant leg up on the capability of the integrated GPU. And that advantage could play very well in the heart of the workstation market.

That's why I not-so-obviously emphasized Llano over Bulldozer. Yes, a Bulldozer Opteron could definitely have a play once again in the workstation market, but it's unlikely to have anywhere near the differentiation over Xeon that Hammer Opteron did over Xeon back in 2004-2005, and that wasn't enough for AMD to put a big dent in the smaller-volume, higher end of the market. So for that (and other reasons), if I were AMD I wouldn't base a re-entry into the market on Bulldozer. Exploit it of course to establish a comprehensive broad set of platform offerings that can address the low to high ends.

But as motivation to get back in, I would try something different, and that "something different" could (or should) be targeted at the heart of the market, mobile and single-socket - especially this emerging "low-entry" category, rather than the niche, low-volume mid-range and high-end. Fusion could differentiate AMD from Intel more effectively in that arena, quite possibly with Llano's superior integrated graphics.

And Fusion (as well as a Bulldozer Opteron + discrete FirePro combo) give the company the chance to pitch workstation OEMs a clean, one-stop-shop solution the company can pitch: a combination of workstation-caliber CPU and graphics, a proposition that neither of its chief rivals Intel and Nvidia can match.

And again, before sticking one's nose up at workstation volumes, compared to broader PC markets, remember the game is about making money, not just shipping the most units. The margins in workstations are far superior, allowing committed vendors like HP, Intel and especially Nvidia to profit handsomely. Just takes some vision, creativity and commitment.

Cheers,
Alex Herrera
Consultant and Jon Peddie Research workstation and professional graphics analyst

[aidanjt]

Thanks for signing up and posting, but it's still the case that workstation and server CPUs are identical, which was my initial point. And I'd already agreed that the motherboards generally have some minor differences. Anyway, hopefully AMD will be able to pull it together and make a determined push back into the workstation market, otherwise Intel will end up being re-entrenched with the OEMs and customers, not a particularly pleasant prospect.