• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Barcelona. Will it be here in June then?

Firegod

Firegod

Firegod

Soldato
Joined
12 Sep 2003
Posts
11,822
Location
Newcastle, UK
I know I know, another AMD Barcelona thread. :o But I was just reading on the Inquirer and they think June. I just wanted to know if there is any truth in this? I'm getting itchy waiting for it!! :)

I did do a search, but it seems most Barcelona threads discuss the spec of the chip and not the release date. :)

Thanks all!
 
June/July for Bercelona Quad Cores and Agena FX (Skt 1207/1207+)
Q3 for the Agena Quad Cores (Skt AM2/AM2+)
Q3 for the Kuma Dual Cores (Skt AM2/AM2+)

Think the names are right, even if not, the dates are definitely correct as of the latest AMD roadmap.
 
How quickly do you think we'll see bios updates available to allow usage of these chips in AM2 boards? Presumably the hardware vendors would rather we upgraded and be pretty slow to release the updates?
 
Emlyn_Dewar said:
How quickly do you think we'll see bios updates available to allow usage of these chips in AM2 boards? Presumably the hardware vendors would rather we upgraded and be pretty slow to release the updates?
Click to expand...
I'd imagine the big vendors would release bios updates fairly quickly like they did with dual core cpu's.
 
Yeah, i would hope that bios updates would be out for the launch, if not within a week or two after.
They have the specs plenty early enough to get the bios sorted for a timely release.
 
caithomas44 said:
the new "quad cores" AMD Athlon 64 FX-74's are officialy here, imports from the states are avalible now :)
Click to expand...

Those are old K8's on a 4x4.

AMD's lastest roadmap for Phenom CPU's shows Q3 for the quad cores (both server and desktop) parts, and Q4 for the dual core parts.
 
Has to be considerably better then Core 2 Duo really, and in all disciplines.

Early rumors indicated that the new AMD's might for instance have 2 FPU's per core. While this is great for the scientific community, for a gamer it would have hardly any impact, as most games still rely heavily on the ALU, and Core2Duo's ALU is extrememly efficient already. Likewise for gamers SSE performance is probably more important than floating point these days.

Im interested in seeing the single threaded performance of the new AMD. That will show us the chips strongpoints best.

If AMD demo Phenom V a Kentsfield, it would be pretty easy to tailor a suite of synthetic benchmarks which highlight the weakness of the FSB link in Kentsfield. But honestly many real world applications dont show the same limitations.

Anyway, if the main gains for the new AMD chip proved to be from the better integration of the processor cores, then intel could release a native quad core part, with their CSI bus (intels take on hypertransport), and that could easily bring them back on top again.

So fingers crossed that AMD truely have a high performance core, and are not just relying on massive power for highly multithreaded applications.

If they do rely heavily on multithreaded performance, intel still have the hyperthreading 'ace' to play, as its strongly rumored that hyperthreading will be put into play again in the not to distant future.

Native Quad core, able to process 2 threads per core anyone? Certainly not impossible, Sun's Niagra processor is able to process 32 concurrent threads on its 8 core processor, obviously that level of parallellism is possible as Sparc processors are pretty 'simple' RISC chips, but I can easily imagine a Core 3 Quad running somewhere between 8 and 16 concurrent theads in the near future.
 
For gamers, C2D's strength is in it's overclockability. Most of us are getting massive gains, even 'bad overclockers' are giving very decent results for the price.

I hope AMds new cpus will offer great performance. Doubt they'll overclock as well though.
 
Corasik said:
Has to be considerably better then Core 2 Duo really, and in all disciplines.

Early rumors indicated that the new AMD's might for instance have 2 FPU's per core. While this is great for the scientific community, for a gamer it would have hardly any impact, as most games still rely heavily on the ALU, and Core2Duo's ALU is extrememly efficient already. Likewise for gamers SSE performance is probably more important than floating point these days.

Im interested in seeing the single threaded performance of the new AMD. That will show us the chips strongpoints best.

If AMD demo Phenom V a Kentsfield, it would be pretty easy to tailor a suite of synthetic benchmarks which highlight the weakness of the FSB link in Kentsfield. But honestly many real world applications dont show the same limitations.

Anyway, if the main gains for the new AMD chip proved to be from the better integration of the processor cores, then intel could release a native quad core part, with their CSI bus (intels take on hypertransport), and that could easily bring them back on top again.

So fingers crossed that AMD truely have a high performance core, and are not just relying on massive power for highly multithreaded applications.

If they do rely heavily on multithreaded performance, intel still have the hyperthreading 'ace' to play, as its strongly rumored that hyperthreading will be put into play again in the not to distant future.

Native Quad core, able to process 2 threads per core anyone? Certainly not impossible, Sun's Niagra processor is able to process 32 concurrent threads on its 8 core processor, obviously that level of parallellism is possible as Sparc processors are pretty 'simple' RISC chips, but I can easily imagine a Core 3 Quad running somewhere between 8 and 16 concurrent theads in the near future.
Click to expand...
I've seen this argument mentioned serveral times on various danish sites discussing processor developments. I only really got interested in the whole hardware market again late last year, and i'm really struggling to see the point in hyperthreading? I'm not sure how it's executed, so i'd really like to hear some details on it, i can imagine 2 things:
1: Simple split of the totalt CPU capacity, which wouldn't in any way be a good solution?
2: Dynamic reallocation of resources to a 2nd thread to match the processing power it actually needs.

Many games are still not taking advantage of Dual Core processors, so the question is how extra threads without extra cores will actually help in this situation? I'm still basing this off games, seeing as most gamers will not be running any CPU intensive programs simultaneously with the games?

Please note that my knowledge of this pretty much equals zero, so this in no way a critique, just a request for more knowledge :)
 
gurusan said:
And you're basing this on what?
Click to expand...

well it might hold some truth, in recent years AMD chips haven't clocked as well as there intel counterparts but always done more IPC until intels core architecture, so if K10 hammers core in IPC clock speed will once again be rendered irrelevant in comparing the two, there was a thread a while back saying 2.5Ghz barcelona outperforms 3Ghz kentsfield, not sure how much truth there was in that thread mind
 
Talrinys said:
I've seen this argument mentioned serveral times on various danish sites discussing processor developments. I only really got interested in the whole hardware market again late last year, and i'm really struggling to see the point in hyperthreading? I'm not sure how it's executed, so i'd really like to hear some details on it, i can imagine 2 things:
1: Simple split of the totalt CPU capacity, which wouldn't in any way be a good solution?
2: Dynamic reallocation of resources to a 2nd thread to match the processing power it actually needs.

Many games are still not taking advantage of Dual Core processors, so the question is how extra threads without extra cores will actually help in this situation? I'm still basing this off games, seeing as most gamers will not be running any CPU intensive programs simultaneously with the games?

Please note that my knowledge of this pretty much equals zero, so this in no way a critique, just a request for more knowledge :)
Click to expand...


If you read what I wrote once more, you'll notice that my main interest was infact how the AMD Phenom will perform in strictly singlethreaded applications (Like games). This is where AMD need to really be strong, because intel's current Quad is not as efficient as a 'native' design, and if all AMD's claims are based around this 'flaw' in Core 2 Quad's design, then as soon as Intel move to native, that advantage is lost.

Also, it needs to be strong in all aspects, ALU, FPU, and SSE. At the moment AMD have a very good FPU, and if its true that Phenom will have two FPU's per core, thats a lot of processing power for the scientific community. However gamers, and multimedia users put a lot more demainds on the ALU, and SSE units. Core 2 is a well balanced chip which does well in all fields. For AMD to really take the crown it needs to be in real applications not synthetics.

As I said before, if AMD Phenom's main performance gains are due to being a native quad, then it would be a prime target for an intel native quad with hyperthreading. If on the other hand Phenom smacks down Core 2 Duo across the board, then intel will need a lot more than a few 'tweaks' to regain the performance crown.

I certainly hope AMD pull a rabbit out of the had, and drop a fantastic processor into the market. Intel V AMD battles help push down the price of performance computing for the masses, and maximum performance for the lucky few in the extreme markets.


As for hyperthreading, well in laymans terms, a CPU has many parts, and often a single program isnt able to make use of all available parts. Hyperthreading allows a second program to use the parts that are idle. A single core P4 for instance is able to issue 3 instructions per clock cycle. But many applications only issue 1-2 instructions at a time, and the 3'rd goes to waste. HT picks up the slack. Its more efficient than simply timesharing the cpu between different threads. Although as you say its only of interest to well written multithreaded applications, or people who run a bunch of apps in parallel.
 
Last edited:
Gashman said:
well it might hold some truth, in recent years AMD chips haven't clocked as well as there intel counterparts but always done more IPC until intels core architecture, so if K10 hammers core in IPC clock speed will once again be rendered irrelevant in comparing the two, there was a thread a while back saying 2.5Ghz barcelona outperforms 3Ghz kentsfield, not sure how much truth there was in that thread mind
Click to expand...

Barcelona is a native quad, with integrated memory controllers, V Kentsfield's FSB linked twin dual core design. Im sure its no problem at all to show a 2.5Ghz Barcelona outperform a 3Ghz Kents in synthetic test, or hand picked applications. What we really need to see is how well the new chips perform running singlethreaded applications, because if all gain is based on being native quad, then intel will have much less work to do before they catch up again.

Not to mention intel are hoping for a 20% gain moving from Conroe to Penryn for 'general' applications and up to 40% gain on Multimedia.
 
I see what you're saying Corasik. Basically the chip needs to have more on it's side than just the advantage of being "native" as that only gives one thing for Intel to improve on (with their chip) before re-gaining the performance lead.

Here's hoping. :)
 
You must log in or register to reply here.
Back
Top Bottom