the NDA has been lifted and here are all the video encoding benchmarks i could find:
what an epic fail.
+ Reply to Thread
Results 1 to 17 of 17
Last edited by deadrats; 12th Oct 2011 at 03:41.
Hey deadrats, thanks for updating your post with all the benchmarks. I check Toms Hardware and Anandtech every day but Now I won't have to search for the others!
somehow intel already knew it. they have been pushing back release dates for faster chips for a while now as they saw no need(i.e. no competition).--
"a lot of people are better dead" - prisoner KSC2-303
Future is beginning to look up for AMD. But atlas while i am an AMD fan. The performance bump of the i72600K that can be had for less than $100 is to good to pass up.Murphy's law taught me everything I know.
But you run 7-zip all day the FX-8150 is the CPU to have! It's probably good for x264 encoding in CRF and QP modes too.
Last edited by jagabo; 12th Oct 2011 at 11:00.
There were rumors about Bulldozer not living up to expectations months ago. Too bad they are true.
Beating Intel would have been nice. but I didn't expect it. However I would have expected the Bulldozer CPU tomshardware tested to at least equal the fastest Phenom II x4 CPU in the test for single-threaded applications. Very disappointing.
Last edited by usually_quiet; 12th Oct 2011 at 11:48. Reason: corrected name of website running the test report I read
Murphy's law taught me everything I know.
there is something to keep in mind; AMD's marketing department is responsible for a lot of misconceptions about Bulldozer. we've been told that each module is composed of two 128 bit integer units (ALU's) mated to a single 256 bit fpu. thanks to the thread scheduler AMD uses each ALU appears as a separate core and AMD has been marketing these cpu's as 4/6/8 core cpu's. this is misleading.
Intel cpu's, since Bloomfield have had three 128 bit integer units per core, all three could do integer based math and one was used for Boolean evaluations (it was a dual purpose unit).
AMD has been talking about bringing hyper-threading to its cpu's for a couple of years, but despite the fact that they have a cross licenses agreement with Intel they wanted a different HT strategy than Intel's, a more efficient strategy.
the Bulldozer is what they came up with. if we view the Bulldozer benchmarks as 6 or 8 core cpu's, then the results are a let down; what we need to remember is that all Phenoms, since the first ones, had two alu's per core, Bulldozer just allows each integer unit to be seen as a separate core.
keeping that in mind if we view the benchmark results as a hyper-threaded dual/tri/quad core cpu, the results no make sense and represent a good improvement over the previous generation of AMD cpu's (this view explains why an "8 core" Bulldozer is barely faster than a hexa core Phenom 2, if the same technology were applied to an X6, then it would look like 12 "cores" with a Bulldozer cpu and the benchmarks would be appropriately faster.
i also have a feeling that were the Bulldozer to be benchmarked with apps that allow more than the default number of threads to be launched, that we would see some solid gains; my e7400 gains some decent fps when i manually change the number of threads from 2 to 3 to 4 using x264 from within avidemux and when using main concept's encoder from within Magix that allows to manually set the thread count, the fastest encodes are done with 5 or 6 threads.
keeping in mind that AMD designed Bulldozer to be a server chip first and foremost, one has to wonder if we wouldn't see different results were the tests to be conducted manually setting the thread count to 16-24 on the "8 core" 8150.
Many existing single-threaded application will never be updated to use new instruction sets, so there is a ton of software in use that will never run as fast on a Bulldozer system as it did with some earlier AMD CPUs.
I have an AMD system, so I am not an AMD basher. I'm just disappointed that performance is not better in this area, since I still use quite a few older single-threaded programs.
I to am disappointed that it is not what i expected, while overall it is indeed an improvement over what they have been offering it's not much better than an i2500K. While costing slightly more, and the price of the i2500 will surely drop when intel introduces new chips in Dec.
Last edited by dragonkeeper; 12th Oct 2011 at 19:25.Murphy's law taught me everything I know.
i sent Anand Lal Shimpi an email asking him if he would run some additional benchmarks with x264 hd 4.0 and cinebench 11.5 but with manually setting the thread count to 16 and 24 threads; i think that Bulldozer may only really shine when the thread count is ramped up, higher than normal cpu's would use.
Last edited by jagabo; 12th Oct 2011 at 20:34.
maybe i just don't want to believe that a processor that was promised as far back as late 2009, got pushed back and features were removed again and again; i guess i'm hoping that there is a silver lining to the cloud, because i have to tell you, looking at prices of processors and motherboards from various sources, i think one would have to be crazy to build a Bulldozer based system over a SB based one, as things currently stand.
Murphy's law taught me everything I know.