Legit Reviews

AMD Phenom Processor Update - Memory Bandwidth

Not much has been going right for AMD in recent weeks, but I'm happy to say one of the most popular benchmarks used in testing processors wasn't ready for Phenom when the media hacked apart the poor processor apart on launch day. It seems SiSoftware Sandra XII was not correctly optimized for the Phenom processors! The latest update of Sisoftware Sandra has the right optimization set and ready to go. See what the difference makes for the memory bandwidth score!

"On Sandra XII SP1 our memory bandwidth was 9624 MB/s on the Integer and 9637 MB/s on the Floating point, which is double what Sandra XII was showing. It is obvious from the results that using one thread per core instead of one thread per package is not the best way to benchmark Phenom. The reason I'm bringing this up is because our article on the Phenom processor contains data in it that is inaccurate and it needs to be updated and clarified."

Article Title: AMD Phenom Processor Update - Memory Bandwidth
Article URL: http://legitreviews.com/article/617/1/

hmm. very interesting. I wonder if it will cause any sites to change their scores?

You need to re-patch the latest patch of the original review.

------------------------------------------------

"On Sandra XII SP1 our memory bandwidth was 9624 MB/s on the Integer and 9637 MB/s on the Floating point, which is double what Sandra XII was showing. It is obvious from the results that using one thread per core instead of one thread per package is not the best way to benchmark Phenom. The reason I'm bringing this up is because our article on the Phenom processor contains data in it that is inaccurate and it needs to be updated and clarified."

------------------------------------------------

SiSoftware has this to say about what happened:

"I... some CPUs get better performance with one thread per core (Yonah, K8 DDR2, Phenom, future Intel CPUs ..."

Good job Apop and LR this is exactly why when I think of review sites I think of LR first! The facts, straight up and on the money, and when a little oops happens we hear about it and get the straight dope on it.

Nein wrote:You need to re-patch the latest patch of the original review.

------------------------------------------------

"On Sandra XII SP1 our memory bandwidth was 9624 MB/s on the Integer and 9637 MB/s on the Floating point, which is double what Sandra XII was showing. It is obvious from the results that using one thread per core instead of one thread per package is not the best way to benchmark Phenom. The reason I'm bringing this up is because our article on the Phenom processor contains data in it that is inaccurate and it needs to be updated and clarified."

------------------------------------------------

SiSoftware has this to say about what happened:

"I... some CPUs get better performance with one thread per core (Yonah, K8 DDR2, Phenom, future Intel CPUs ..."

The benchmark in the original review has been patched and benchmarked again already... I did that at the same time I posted this article. If you mean to benchmark each processor with the same technique that's not right as every benchmark has to be optimized for the system they run on. Sandra has always been like this.

Good job on the update! I don't see other sites doing it (so far).

It's nice to know the Phenoms suck a little bit less

so when does the 9900 come out? those are some impressive numbers(as opposed to the failure we thought it would be doomed to )

skierkid450 wrote:so when does the 9900 come out? those are some impressive numbers(as opposed to the failure we thought it would be doomed to )

no date set in stone yet... They are waiting for a new stepping I understand and once those get kicked out the door and yields are good enough you will see the 9900 hit the market... trust me if AMD could release a 9900 now they would as they need the performance boost.

Apoptosis wrote:The benchmark in the original review has been patched and benchmarked again already... I did that at the same time I posted this article. If you mean to benchmark each processor with the same technique that's not right as every benchmark has to be optimized for the system they run on. Sandra has always been like this.

I meant "Pentium D, Merom, Penryn, etc.." which are Symmetric Processing designs and only good for one thread at a time, or ALA Hyper-Threading = pretend to be Multi-Threading with single big pre-optimized thread.

As opposed to "Yonah, K8 DDR2, Phenom, future Intel CPUs" which are Distributed Processing designs and most optimum with actual Multi-Threading AKA one thread per core.

What nein means is that you wrote this line

It is obvious from the results that using one thread per core instead of one thread per package is not the best way to benchmark Phenom.

the wrong way round.

The new sandra patch uses one thread per core to get the better results, not one thread per package like you said above.

SiSoftware SANDRA wrote:If you check the “benchmark breakdown” you see it is now using one thread per core instead of one thread per package (physical CPU) like earlier versions did.

Just delete the "not" and your set.

Dan

Ahh, much better! Nice to see AMD back at the top of memory bandwidth benchmarks atleast