So any chance we'll see 1 socket C32 workstation boards with OC features?
Printable View
So any chance we'll see 1 socket C32 workstation boards with OC features?
Asus we'll probably do it. After the TUF series I have little doubt. I would prefer someone making a 2 socket system that can be overclocked instead.
As JF-AMD said, most likely no.
Well that's quite understandable. I'm talking about base clock OC features here. I'm not even sure AMD can do anything if e.g. ASUS releases a board with a bios that allows the user to adjust the base clock. I dont expect AMD to release any unlocked Opteron chips and if they do they'd be FX chips anyway.
While I understand that AMD might not be terribly concerned with the ultra high end market, why not at least encourage your board partners to make some enthusiast friendly c32 boards? Why just outright hand over that segment to intel when you will have 12 and 16 core bulldozers that I'm sure will be competitive with s2011 chips.
Corporate business practice and stiffness of big old organization come to mind.
But i believe that if bulldozer will be good, they might release some renamed server parts, unlock them ,and charge 1k a pop .FX51 comes to mind.
Actually, it is a money losing proposition. When the world was dual core, 2P client products were ~.8% of the market. Then when quad core came out, that dropped to ~.4% of the market. It has continued to go down and with 8-core around the corner, it is probably getting pretty close to 0%.
When you compete for internal resources on a project you have to justify the business case. A new part is typically ~$5M to get on the roadmap.
I don't make the decisions on where to invest on the client side, but if it was up to me, spending $5M to bring in a few hudred thousand of revenue is not a good business plan.
Im not sure with which part you disagree.
That AMD wont bring FX back (if it has performance)
That AMD wont bring FX back in c32 format (by just enabling OC and multiplier unlock on dedicated cpu/mobo part)
Or that AMD wont bring another 4x4 (2P part with some unregistered ram etc).
With the last one it seems vaguely possible it could cost 5M.But with the first two, theres no reason it should cost more than bringing any other cpu part.
And i really hope it doesnt cost 5M to come with athlon 260 if 255 is already
on the market.
One new high margin cpu sku, and a green light to mb makers.
On a side note, if hypothetically, asus would come to amd with a question if they can build OC capable mobo for c32 platform.Would AMD in any form discourage that action ?
Theres two sides to this.
JF already said many times, that they dont allow OC options because of the alleged influence on the brand.That it makes it look "unprofessional".
So AMD may be actively discouraging partners from doing that.
Besides, were talking abouta future product ,which may change the situation in the "worth investing" category.
There were such products before on both camps.
And on the intel side, theres even one now.
For the few last years, AMD just wasnt competetive enough with its products in this segment.
But i hope bulldozer will change that.
Besides, theres more to such products than money only.
Although i doubt intel loses on 980x sales, or evga on SR2 sales.Or asus on ares sales and so on.
If there is going to be FX part back,it will be single socket part.The market for 2P enthusiast system is really too small to justify such kind of investment.
On the other hand,if you have money like intel does and you are crazy enough,you can make such a board (and CPU models),even if it means you will actually lose money on the project in the end.All for the "wow" effect.
On the 1P part, AMD could still do a socket 940(server part renamed).So it would get 16 core chip,dual dies.Instead of a simple rehash of the desktop part.
I mean, FX part should differ somehow from a black edition cpu.What else is there beside more cores ?200mhz speed bump ?
With 4x4 AMD at least tried ;-) .And it would be a success if they would have competetive architecture.
16 core would require a larger package. We won't be sharing infrastructure between server and client, it just doesn't make sense based on costs and revenue opportunity.
double post (all the way)...
c32 would bring almost no added value in terms of performance vs am3 besides the capacity to use 2p. g34 is what is interesting to me and IF amd ever decided to compete head on with the s2011 sb from intel it would have to be with a desktop 12 core (16 is too much and would be too hot in high clocks) with quad channel memory. That wiuld be worthy of the FX name and it would definately allow for board makers to add oc features to the g34 platform. But as someone already said, amd believes that deteriorates their professional server image so it wont happen anyway.
Wishing for c32 is not much of an advantage. I would much rather see single g34 phenoms around.
Well all we can do is hope that BD is great and we wont need to overclock anything. Supermicro and Tyan will have (already have boards too) that will fit most fine. And you can use unbuffered ram in them already if you stay under i think 4 modules per CPU
Thats a pipe dream im afraid.s2011 from intel is going to be overclockable through multis and blck.Plus its pretty much sure thing it will be clocked quite high on the get go, with a solid mainboards to boot.
Hard to imagine G34 opteron on default would be comparable to 8 core highly overclocked sandy bridge using unbuffered ram on 4 channels.
With current management attitude that battle is almost certainly lost already.
AMD seems to forget about performance workstation market, and wont have a chance for powermac`s too.Its a pity.
I firmly believe that architecture itself is going to be competetive, but there wont be a platform that can utilize it.
What are BD chips?
A bios update may update the chipset to be able to use DDR3 1600mhz but without SPD settings, it won't be set unless there's a way to set it (aka overclocking settings in desktop mobos)
Solution!
FOUND IT!
http://img.photobucket.com/albums/v1...elixircpuz.jpg