AMD Bulldozer Core Patent Diagrams

hoom said:
Worrying really.
Conroe engineering samples were giving out very promising scores early.

Either they made a big booboo somehow that will make the TLB error look like a minor oops or these 6 core engineering samples are severely crippled.
Click to expand...
Indeed if ther si truth to these captures it's really shaping badly :devilish:
It somehow reinforce my belief that AMD should have stop competing for the high end with Intel and focus on mid and low power CPU even though the last news about about Intel using tri-gate transistors on their 22nm process would have make this tougher than than expected.
AMD can't afford to work on bobcats, bulldozer, llano, etc, they should be more focused and it shows. I've the clear feeling that the rebound in their income is not meant to last (or either Intel purposely makes it lasts).
 
liolio said:
Indeed if ther si truth to these captures it's really shaping badly :devilish:
It somehow reinforce my belief that AMD should have stop competing for the high end with Intel and focus on mid and low power CPU even though the last news about about Intel using tri-gate transistors on their 22nm process would have make this tougher than than expected.
AMD can't afford to work on bobcats, bulldozer, llano, etc, they should be more focused and it shows. I've the clear feeling that the rebound in their income is not meant to last (or either Intel purposely makes it lasts).
Click to expand...

Yeah Intel needs to take another dive, P4 style. It's better for everyone ultimately.
 
are there scheduling woes? back then on the core i7 920's release there were stupid results with hyperthreading enabled. depending on the windows version, threads were not being numbered the same, so thread 0 and thread 1 would run on two different cores rather that two threads of a core.

the other issue would be, no single thread improvement :p
 
Those clocks in Felixs' last shot look rather promising :smile:

2.8Ghz base clock with 1Ghz turbo to 3.8.
Presumably if HT is 3.2 then northbridge/L3 are too?
And within a 95W TDP.

Can't speak for those benchmark results though, good/bad?

If that one is true though someone must be looking pretty fired around now :oops:
 
hoom said:
Can't speak for those benchmark results though, good/bad?
Click to expand...
That Super PI result in the last shot is ~3.5 times faster than the ones in the earlier batch of results from Chiphell. It's also much faster than any Phenom II.
 
The screen-shot is fake. At least so says Gigabyte - read the update.
There are two windows in there at focus -- the chat application and the second CPU-Z instance in the right. Something has been added later for sure. In the left CPU-Z window, the cache info section does not indicate the number of caches, according to the number of cores/modules.
 
Well sure Gigabyte is hardly going to publicly say "Oops, yeah one of our staff put a screenshot of AMDs NDAed top-secret upcoming processor details on the net" are they?
Of course, if it was legit the update would have probably been a lawyers' takedown notice rather than a 'its fake'.
 
Many errors in the Gigabyte pic, it is definitely fake.
The scores would also make it a lot faster in Super Pi than the 2600K.

Example of errors, no MSN in taskbar and multiple active Windows (notepad, CPUz and MSN chat window).

If it was real you could have expected Bulldozer to demolish Sandy Bridge.
 
Tahir2 said:
Many errors in the Gigabyte pic, it is definitely fake.
The scores would also make it a lot faster in Super Pi than the 2600K.

Example of errors, no MSN in taskbar and multiple active Windows (notepad, CPUz and MSN chat window).

If it was real you could have expected Bulldozer to demolish Sandy Bridge.
Click to expand...

isn't super pi interger units not the fp units ? if so a 3 modual bull dozer would equate into 6 interger units. I believe sandy bridge has 1 interger unit per core so it would only have 4 interger units.
 
eastmen said:
isn't super pi interger units not the fp units ? if so a 3 modual bull dozer would equate into 6 interger units. I believe sandy bridge has 1 interger unit per core so it would only have 4 interger units.
Click to expand...
Super Pi is single-threaded, and as of now, one Sandy Bridge core is ~2.2X faster than a current AMD core at the same clock speed in this old benchmark.
 
That Super PI is very usefull for everyone who wants to calculate PI number for the whole day. :p
How about leak more usefull benchmarks next time. Some memory, cache benchmarks ,PCmark or some games finaly. But i guess faking Super PI numbers is much easyer.
 
super pi is great at getting a quick stability result, it catches errors right away on systems that are stable enough for most desktop use and some gaming.
 
If you guys remember way back even Opteron had performance issues prior to launch so don't get disappointed unless they launch a wreck. If they do launch a wreck, the company might just be doomed though. I don't really see how they can continue to sell huge dies for super cheap. You don't make back your R&D that way.

BD is supposed to be designed to rock the server world most so hopefully it does that...
 
lol at that super pi screenshot. It really is fake. It has 2 recycling bins. one on the upper right, one one the lower left.
 
av8fb.jpg


p.s.: Source
 
Last edited by a moderator:
Is the CPU score formed solely by the CPU tests or there is also some of the GPU score factored in? Because a quick google search for vantage scores shows people with higher CPU score although lower results on the CPU tests. Also it's weird that AMD would supply him with an ES but won't bother to provide an AM3+ motherboard (the one he is using is AM3).
 
You must log in or register to reply here.
Back
Top