I really wasted probably days of my life reading through all this speculation. I think I should make a change on how I spend my time
...
I really wasted probably days of my life reading through all this speculation. I think I should make a change on how I spend my time
You registered five years ago and just made your first two posts?
Is that a record?
The funny thing is that after 1900 posts in this thread I don't really know anymore about Bulldozer than I did before the thread started. I really wasted probably days of my life reading through all this speculation. I think I should make a change on how I spend my time
BD might have a separate immediate/displacement const unit and one related patent mentioned that it might store 4 32b values in its registers or combine them to store two 64b values. Open64/GCC BD related code dispatch optimization code suggests, that this might really be the case.AMD processors have always been able to decode "enough" immediates, because the decoders are symmetric.
You registered five years ago and just made your first two posts?
Is that a record?
How did he even know about Socket F in 2006? Perhaps he is an engineer.
Socket F was announced in 2005 introduced in 2006?
Socket F is a CPU socket designed by AMD for its Opteron line of CPUs released on August 15, 2006[1]. In 2010 Socket F was replaced by Socket C32 for entry-level servers and Socket G34 for high-end servers.
Are you people really that bored?
This thread simply will not give up the ghost.
Possible FX 8 core processors results leaked.
Possible FX 8 core processors results leaked.
This chart has been posted 3 times in this thread already. The majority of the last 3-4 pages were talking about these results and how likely they were to be real or not.
This chart has been posted 3 times in this thread already. The majority of the last 3-4 pages were talking about these results and how likely they were to be real or not.
According to that graph (as I understand it) then the FX-8 core is only 30% faster than a 6 core PhII? If that bench is up to 8 threads, then BD looks blah.
If the bench is up to 4 threads only, then BD looks blah since the turbo clockspeeds are supposed to be much higher than PhII turbo. So AMD will be counting on much higher clockspeeds just to beat their previous flag-ships? :thumbsdown:
So at best that bench is only showing a 30% increase over their previous flagship cpu? Hopefully I don't understand the graph and/or the graph is a fake.
Jason
As I stated in an earlier post, the even more disturbing metric is that the fastest Llano processor (The A8) is 30% slower than the fastest Thuban. It also shows it being 70% slower than both the i7-2600k and the fastest BD processor rated.
These measures really call into question the validity of this graph, as they just don't make much sense.
Even if they are/were 100% legit results, the chances of them being reflective of retail/shipping motherboard BIOSes and CPU steppings (including requisite microcode) rapidly approaches zero with every day that stands between the date of the bench leak and the date of retail availability of the CPU/platform on Newegg.
Leaked benchmarks, being the static snapshots they are of a very dynamic iterative optimization cycle that transpires right before a product launch, are worse than worthless IMO as they are virtually assured of being irrelevant to the performance of the product we ultimately have the opportunity to purchase whenever it is finally released.
(this is even more apropos when GPU elements are involved, performance is moving target based on monthly driver releases even after the hardware goes retail)