FX6 and FX8 have more cores than 2600k tough.Quote:
Originally Posted by Musho
Printable View
FX6 and FX8 have more cores than 2600k tough.Quote:
Originally Posted by Musho
Core count doesn't really matter in terms of pricing. Performance has historically been used so long as they aren't cutting dangerously low to die cost.
FX6 has 3 modules and 6 threads, FX8 has 4 modules and 8 threads, while the i7 2600 has 4 cores and 8 threads. what are u trying to say?
the high price for the 4110 suggests that it does very well with turbo to push the limits of the power envelope
basically for 95W you can get 6 fast cores or 4 really fast cores, otherwise we would have seen it priced way lower than the 6110
1.272v for full load with turbo on!!!!
btw you might want to be careful with that, we can measure the pixel count and see how long the extra ram was used to determine how long the test has been running,
and what % is left and get a pretty good estimate of the total time it would take to finish, then convert to a score
lol :rotf:Quote:
Originally Posted by manicdan
@liberato87: that is a FAKE. and no, im not just thinking it could be, i KNOW it is.
I never said it is official (nobody can say that). IMHO it is the screen nearest to the reality (other screens : vcore too high, frequency too low )Quote:
Originally Posted by w0mbat
also in these days we ve seen some photos hosted on that site with crosshair V and AMD "strange" boxes. so I think it maybe true!
Tell us what do you know about that! thanks!
this is a complete photoshop, it was meant as a joke. i know the guy who faked it from a german forum.
I was wonder How they got firefox before the start quick start icons.Quote:
Originally Posted by w0mbat
by the way bulldozers has already been noted to have a 2.4ghz North bridge speed.
so I take it we all believe it has a 3.2ghz Hyper transport?
And 2600K is on average just ~4-5% slower than 980x while being 4x or more cheaper. Meaning cores do little on desktop,except in few select applications. Microacrhitecture,how it behaves in real world workloads and actual support in applications all play a major role.Quote:
Originally Posted by Mechanical Man
source?Quote:
Originally Posted by w0mbat
i dont know how to recognize a fake but
http://errorlevelanalysis.com/permalink/9440e01/
and this is my screen analyezed
http://errorlevelanalysis.com/permalink/e34fb4d/
i think the result is the same.. and I not need to do a fake for 7.12 cinebench lol
That's normal, they aren't quick start just pinned.Quote:
Originally Posted by demonkevy666
Image posted by liberato87 is a 100% fake. Because it is already known that in CPU-Z string HT-Link/Rated FSB for Bulldozer must be EMPTY. Author of CPU-Z have posted the real CPU-Z image for Bulldozer in his article: http://www.hardware.fr/articles/833-...bulldozer.html
Also it was confirmed by one of the testers (who already received Bulldozer) in his blog. Unfortunately he already deleted the entry, which had some additional info about Bulldozer memory controller speed and power consumption.
Any screenshot that has CPU-Z showing some kind of HT-link frequency - is fake.
PS For the abovementioned screenshot - also please look at how the model's right hand can be seen through Windows Task Manager window. It's clearly shifted to add more cores. ;)
i AM the source.Quote:
Originally Posted by liberato87
On the server side shanghai cpu's already had an higher HT than NB speed. Where was this 2.4GHz NB speed mentioned? Must have missed that.Quote:
Originally Posted by demonkevy666
A module is not analogous to a core with hyperthreading.Quote:
Originally Posted by w0mbat
If it has a 2.4 Ghz NB speed then it will have a 2.4 Ghz HT speed...Quote:
Originally Posted by demonkevy666
HT 3.0 as 890FX boards have currently is good for 5.2 Ghz or 2600 stock...
A module is exactly analogous to a core with hyperthreading; if they price it the chips in the same price range.Quote:
Originally Posted by Particle
If they don't want a module to compete with a hyperthreaded core, then they will price their chips accordingly. Otherwise w0mbat was completely correct.
But then you were also correct in that performance and pricing is more paramount anyway. When people point to an Intel part that is 4x the price and 15% faster they are not looking at a real world solution.
I guess until we see actual performance AND prices we won't really know much. (I'd love to see the 2M/4T BD compete just fine against a 4C/8T Intel chip... but I would expect more to see the 2M/4T chip competing more with the Intel 4C/4T chip. I guess we'll all probably know in about a month.)
IPC is not the same between any generation or any company
back when cpus were all one core, the GHZ didnt matter since the time when AMD was ahead, they did it with lower clocks. (2500+ running at 1.8ghz for example)
prices are set 90% of the time, by performance relative to current offerings of the total chips features, which has now expanded to single threaded, few threads, and highly multi-threaded performance
No...Quote:
Originally Posted by keithlm
Module is just a term to explain how AMD shares parts between physical cores.
One module has TWO physical cores...HyperThreaded threads have nothing to do with TWO PHYSICAL CORES
Of course AMD can price an 8 core to SB's 4/8...but wombat wasnt talking about competing, he was stating that basically thats what AMD's modules are.
It's almost like buying a pack of batteries in twos instead of buying individual packs of batteries with just one battery inside...you save some space the packaging takes up. That's all a module does.
8 cores, 8 threads. There is no option to turn "HT" off, and single threaded workloads are not going to skyrocket in performance like they would if you were running 4 threads on a 4/8 SB chip rather than 8.
For example if you ran 4 threads of HyperPi on Bulldozer vs 8 for the most part all threads 4 or 8 will perform equally. If you ran 4 threads of HyperPi on SB it would be significantly faster than if you ran 8 threads of HyperPi.
Actually the ONLY thing that needs to be considered when wondering if a module is analogous to a hyperthreaded core is the intentions of the architects for both companies.Quote:
Originally Posted by BeepBeep2
If they were creating their design for basically the same purpose then the designs are exactly analogous.
The questions of whether one design is more elegant or performs better or costs more are actually secondary. (And your last example means absolutely nothing; although is does display the reason Hyperthreading is not an optimal solution.)
Module design intentions: Save die space vs conventional core layoutQuote:
Originally Posted by keithlm
HT design intentions: Make use of unused CPU cycles, because of stupidly long instruction pipelines
Two can play that game:Quote:
Originally Posted by BeepBeep2
Module design intentions: Increase throughput for the amount of space used in the die.
HT design intentions: Increase throughput for the amount of space used in the die.
(And the use of stupidly long instruction pipelines has always been a bad way to increase performance anyway; if you accept your "intention" as being true you are also saying that hyperthreading was created to resolve a poor design.)