Last message on previous page:
So what if AMD fixed the bug. The bug isnt causing problems for the people that use the chip.

What B3 needs to do is clock higher.

If it all does is fix the bug then its a HUGE dissaster. At least get the orginal launch clockspeed cpus out. But they really need to get higher clocked models out.

Agreed. What I don't get is why their video cards are made at 55 nm but the CPUs aren't. That could help with the clocks and power consumption. Isn't ATI's technology available to the CPU division???

But you will be able to upgrade with whatever comes out between now and then. And by the time 2010 comes around, there will still be C2D based 45nm quads around. While its not good to guess, unlike the gamble of Phenom, I feel safe that whatever those 45nm C2Qs are, they should be faster, cooler and down in price.

Good point. Maybe I'll get a Q9550 for $100

Is Phenom faster than X2? Please tell me AMD did NOT just release something slower than they already had, did they???

It is faster, clock for clock, but the B2 retail releases dont clock as high as either the brisbanes or the 90nm, so for the time being, it cant keep up in single threaded (i.e the vast, vast majority) of apps

Wow

Following this chart dual cores is better than Quads

How come ??

AMD got 3200 FBS intel got 1066 Mhz only

AMD got 3 levels of cache Inel got 2 only and level 2 almost do all the job

How AMD phenom become slower ???

AMD (phenom) and Intel is two different hardwareplatforms, when comparing these two you need to know what you are comparing.

Running singlethreaded applications on Intel will be faster because Intel isn’t optimized to run multiple threads. AMD is a much more complex processor, it is designed to scale well when all cores are used. The processor is also designed to have more speed to other parts on the motherboard (memory, gpu, etc).
L3 cache is slower than L2 cache since L3 cache is more complex. L3 cache can synchronize cache among four cores, Intel needs to synchronize cache using the memory and goes through the slow FSB.
Intel has only one connection to all other parts on the motherboard. AMD has one for the memory (IMC) and one for I/O (hypertransport). Also the L3 cache is used to manage cache choherency.

You don't need to overclock AMD because it scales well and it doesn't have the bottleneck that Intel has, the FSB on intel needs to be clocked if you want to maximize performance.

Buying an AMD phenom will probably last for many years, buying a Intel and new applications that are heavily threaded needs a new processor.

My vote is definitely for Q6600, or some other Intel Quad.

Oh my, we got another thunderpants in here.

Or perhaps somebody that just appreciates the truth.

ATI's GPU's aren't fabbed by AMD, they are fabbed by TSMC (Taiwan Semiconductor Manufacturing Company) just like Nvidia's.

Sorry dude but there are only a few things that the HTT and IMS allow Phenom to be better. Thats memory bandwidt benchmarks and server apps.

As for the scaling, it scales better in MP servers not multi-tasking. And normally you will only see the benefit of it in higher than 2 CPU systems.

As far as I have seen, most multi cored benchmarks still show a C2Q besting a Phenom. Of course the memory sensative ones such as WinRar do better on a Phenom but thats still due to AMDs better memory structure of the HTT.

Of course that will all change with Nehalem being introdouced later this year and in 2009.

If you know how Intel and AMD works you can tell in advance how they are going to perform.
If the application can take advantage of the fast and big L2 cache on Intel then Intel will perform very well. If threads doesn’t need to be synchronized then Intel also can execute fast.
Type of applications that will fit Intel is singe threaded games, those games love cache and the FSB isn’t a problem there. Rendering applications, these types of applications does calculate a lot and it is a type of operations that is controlled. Advanced compression; like video compression. These do some intense calculation.
That means that the developer can plan the threads and doesn’t need to check memory and do synchronization more than when the task is finished. Synchronizations is something that all programmers tries to avoid because even if the processor can handle it well it will bring some performance penalty. Also when they test applications there are often so that the applications can run alone on the computer. Try to start some applications that is working and run the same test, you will see that Intel looses its edge pretty fast.

I am not saying that the one is bad and the other is good, I say that they have different targets. AMD is built to scale when the load gets heavier. Intel is built to run small or singe threaded applications as fast as possible.

Bringin' back a 'thread from the dead'? - lol

The Phenoms show some promise in parallel multithreading but won't overclock like a q6600. This is the last month that the 125w Phenoms can be ordered by distributors and by Q4 95w 9850BEs will be rolling out (and hopefully 45nm).

Hopefully soon the 790gx / SB750 mobos will be available and those slackers at AMD can get a bit of their mojo back