My question is do we have anyone here that works for amd?
lol
Printable View
My question is do we have anyone here that works for amd?
lol
I have no doubts it will do 4 gig on air MAYBE but
THAT gave me a BIG laugh - do they even read what they are writing?Quote:
There's said to be two chips in the pipeline, the FX-80 at 4Ghz and the FX-82 at 4.4GHz.
I think deneb will bring AMDs scores a good step further, maybe 100k+ 03 with 4x4850 or sth. like that. I am hopin' to be able to bench it anytime soon so that I can find out myself what it really does when going with it XS-style :cool:
Those clocks aren't plausible even if Buddha says a prayer for AMD.
I'm shocked at how intelligent folks drift into a fantasy for whatever reasons they might have.
K10 was designed to operate at 2.2-2.8GHz@1.15V in a 95w TDP envelope at 65nm.They are at 2.6GHz@1.3V within a 125w TDP.
Obviously something is wrong.
AMD's 45nm ( at least the 1st iteration ) brings absolutely nothing new.As a rule of thumb , you get 20% ceteris paribus.Gains on the order of 1.4-1.8GHz or 50-70% :shocked: are utter BS.Never happened , never will.
Apply to their original goal and you get 3.4GHz@1.15V in a 95w TDP.Apply to their current situation and you get 3.1GHz@1.2V in a 125w TDP.The truth is somewhere in between ; if I were to guess , I'd say 3.2GHz@1.2V@95w.
That is already considered by some very optimistic.Why ?
K10 pipeline is basically the K8 one which does more work at every stage.In other words , it will clock worse than a K8 which wasn't that great in the 1st place.Nothings in the design points out to a speed racer , just decent speed , good performance with minimal fragility wrt SW.
Having said that , original rumors on Deneb were 2.8GHz at launch ; that I can believe. 4/4.4GHz is pure FUD , industrial grade.
AMD bought intels netburst technology?. Nah its just BS, besides the included heatsink couldent be anything less then a TRUE or a small WC kit lol. The powerdraw would be 200-300w.
4ghz quad at 1.16 volts...prolly more 1.45Vcore we have seen before... nice 5min paintjob prolly... AMD would kill it's own lineup unless the FX are up to the old FX price standards...and think no stock heatsink will be able to cool that baby
I think this is a fake and this is why:
:rolleyes: :p: "In the galaxy"Quote:
Originally Posted by reviewage.co.uk
It's taken straight from that link.
This year i think are possible Denebs to 2.8GHz and FX series next year (i mean,defautl clock 3-3.2GHz). OC up 4Ghz is possible, but with more voltage. This is C1 revision, from screen? Hardspell preview was only C0 revision...
I like megatasking more
This reminds of the same BS before Phenom was launched. They promise the world and deliver a cup of dirt. Anyone remember them claiming that Phenom was scoring 30k in 06 at 3Ghz with only 2 HD2900XT's? This is just about as far fetched as that nonsense was.
Chances of this being true = South of 0%
First of all, the 4GHz Deneb screenshot was shown a while ago by someone claiming to have a Revision 2 Deneb sample and overclocked it to 4GHz. Whether or not the screenshot was fake is debatable, but it has nothing to do with a Phenom FX clocked at 4GHz stock. And the CPU-Z voltage is wrong, it is not 1.168V, the screenshot also included an AOD shot showing the real voltage, 1.475V.
That's not even taking into account that 4GHz is impossible to reach @ stock on a quad-core CPU at 45nm. Intel can't do it, there is no way AMD can do it.
The only way AMD gets 4GHz on 45nm is a complete redesign of the core, which we know that Deneb is not. The only way AMD hits 4GHz stock is through a lower IPC (Netburst-like) architecture.
40% better than Xeon "across a wide range of apps" was said by Randy Allen.
Result : they achieved that in 2 tests ; Stream and SpecFP_rate.BS masters.
Look from this viewpoint : I'm amazed they managed to equal Penryn considering the cache hierarchy.
Penryn is 4 issues wide core,excellent INT and FP, 6MB 15 cycles L2.
Nehalem build upon Penryn core , more instructions in flight , buffers , 256KB 11 cycle L2 + 8MB L3 with 39 cycles.
Yes , it's got an IMC , but the advanced prefetchers pretty much cancel that advantage.
It's a very close call , but I'd say there will be some cases where Penryn is faster.That's not to say Nehalem isn't awesome , just that Penryn set the bar incredibly high.
L3 on a 2.66GHz Nehalem is approximately ~50% faster than the L3 on a Phenom 9950 and is 4x the size. L3 on a 3.2GHz Nehalem would be ~70-80% faster than on a 9950. This takes into account frequency (Phenom 9850/9950 have L3 @ 2.0GHz, Nehalem L3 runs @ core speed) and the lower latency of the Nehalem L3.
please see my post here
http://www.xtremesystems.org/forums/...5&postcount=25
could be wrong however
Deneb @ 2.3 ghz vs phenom 9650, deneb uses half less power than 9650 phenom.... theyre allready maded quite drop in consumption... was it at hardspell ?