AMD FX "Bulldozer" Review - (4) !exclusive! Excuse for 1-Threaded Perf.

**BeepBeep2** · 10-12-2011, 05:26 PM

Originally Posted by chew*

good question, one I don't have an answer for at this point in time.

Thank you, that's all I needed to know.

The CPU would be worth the $250 to me had they figured that rather important bit out. Nor would the CPU be portrayed as sucking so badly.

Originally Posted by Baam

So if i disabled 4 clusters, will it run games better?

Yeah, at least I would think so.

I'd say quite a bit better, near i3 2100 level at stock? (Not trolling here, just looking at real data)

chew* · 10-12-2011, 05:30 PM

Originally Posted by Baam

So if i disabled 4 clusters, will it run games better?

Providing the game can not make the use and or benefit from more than 4 threads yes.

**radaja** · 10-12-2011, 05:37 PM

good info,its just a shame to have to buy a FX-8150 and run it as a quad-core to get the performance out of it.
if i can afford it i will do it though,i had planed on a FX-4170 but that wouldn't be a good idea using the disabling
of the b clusters,then i would only have a dual core

maybe i will win a FX-8150 i have netered several contests

chew* · 10-12-2011, 06:31 PM

Originally Posted by radaja

good info,its just a shame to have to buy a FX-8150 and run it as a quad-core to get the performance out of it.
if i can afford it i will do it though,i had planed on a FX-4170 but that wouldn't be a good idea using the disabling
of the b clusters,then i would only have a dual core

maybe i will win a FX-8150 i have netered several contests

Why is it a shame to run it at it's native core count

it's a 4 core with cmt, your just disabling 50% of CMT

Isbre · 10-12-2011, 06:49 PM

Would be real interesting to see some comparison of deneb 4C vs Zambezi 4CU/4C @ same clocks

**BeepBeep2** · 10-12-2011, 06:57 PM

Originally Posted by chew*

Why is it a shame to run it at it's native core count

it's a 4 core with cmt, your just disabling 50% of CMT

Single Thread performance is really important in a desktop environment, thats why.

I really would have liked to see the 12 core Interlagos dies go as 6 core desktop parts minus the CMT @ 4.2 stock but that will never happen and it is much too late for that. Then there would be the argument of "that wouldn't be efficient due to CMT"...

Though power consumption would be down much more so than if CMT was enabled. That would give the strong single thread perf needed as well as multi-thread, or maybe they could just figure out how to disable the shared resources correctly...

**Manicdan** · 10-12-2011, 06:59 PM

+1 for testing 4CU/4C in gaming.

**stangracin3** · 10-12-2011, 07:01 PM

Originally Posted by Manicdan

+1 for testing 4CU/4C in gaming.

+2

so far this has been the best review ive seen.

i want to see MM's wcg numbers also if he would come out of his hole.

**Spectrobozo** · 10-12-2011, 07:05 PM

Originally Posted by Manicdan

+1 for testing 4CU/4C in gaming.

agreed, testing more single thread softwares and some gaming would be interesting,
4m/4t vs 4m/8t,

**tbone8ty** · 10-12-2011, 07:14 PM

hmm interesting...this needs to be tested more, disabling a combination of cluster units and threads in bios and how it performs....

just curious how it performs vs phenom I 9850/9950 vs phenom II

how much did Phenom II increase over Phenom I? was it the 10-15% AMD is talking about with Piledriver?

**Manicdan** · 10-12-2011, 07:24 PM

Originally Posted by tbone8ty

hmm interesting...this needs to be tested more, disabling a combination of cluster units and threads in bios and how it performs....

just curious how it performs vs phenom I 9850/9950 vs phenom II

how much did Phenom II increase over Phenom I? was it the 10-15% AMD is talking about with Piledriver?

PI to PII was near identical except extra L3 i thought, so scaling was better and it clocked better, but IPC i think was very close

**Lokinhow** · 10-12-2011, 07:52 PM

Very, very interesting.

Can you please run low threaded applications (like games) on it, please?

**demonkevy666** · 10-12-2011, 08:07 PM

Originally Posted by Manicdan

PI to PII was near identical except extra L3 i thought, so scaling was better and it clocked better, but IPC i think was very close

O_o http://www.anandtech.com/show/2702/9 blast from the past.

**STEvil** · 10-12-2011, 08:32 PM

ah, finally someone does something above and beyond the rest of the pack

**Llew** · 10-12-2011, 08:40 PM

Originally Posted by DGLee

But there's a flaw: current version of CH5F BIOS doesn's support core-by-core on/off. Am I finished then?

...Not really. Newest version of BIOS supports such function!

Which BIOS version are you using now to gain access to the individual core control? I'm running 0813 and don't seem to have it.

**LightSpeed** · 10-12-2011, 08:44 PM

If it increases the OC ceiling and decreases power consumption it will indeed be very interesting.

I hope future derivatives can greatly reduce the performance hit of sharing the resources. Thats what seems to hold BD back at the moment

**Nintendork** · 10-12-2011, 08:44 PM

A test with the dolphin emulator vs PII will make my day.

Awesome tiny review

**mav2000** · 10-12-2011, 08:57 PM

This is a great great test. Maybe what chew* says is right. Shut down the extra cores and run it as a quad, for lower power and better single thread performance?

Would you also run super pi using the same method...curious to see the result. Also if you can add the specs of what you are running, I dont mind running a 1090t against those results.

**Cecil** · 10-12-2011, 11:04 PM

Im about to pull the trigger on the 8120 and 990FXA-UD3. Does anyone know for sure if it has individual core locking like this?

**Marc HFR** · 10-13-2011, 12:06 AM

Originally Posted by Manicdan

+1 for testing 4CU/4C in gaming.

http://www.hardware.fr/medias/photos...IMG0033832.gif

Source : http://www.hardware.fr/articles/842-...acite-cmt.html (in french, sorry)

**Spectrobozo** · 10-13-2011, 12:22 AM

Originally Posted by Marc HFR

http://www.hardware.fr/medias/photos...IMG0033832.gif

Source : http://www.hardware.fr/articles/842-...acite-cmt.html (in french, sorry)

thanks,
I think this shows that CMT works quite well (considering how well the 2m with CMT looks against 4m without CMT), and the benefit of disabling CMT gives a limited improvement in some cases, I don't think it worth loosing the extra 4 cores for this gain,
and this is going to be slightly improved on Windows 8 anyway, I think

**The Coolest** · 10-13-2011, 01:16 AM

It's interesting to see the effect on single threaded benchmarks, like cinebench and superpi when cluster B is disable for comparison.

**Spectrobozo** · 10-13-2011, 01:45 AM

Originally Posted by The Coolest

It's interesting to see the effect on single threaded benchmarks, like cinebench and superpi when cluster B is disable for comparison.

many of the games tested use less than 4 cores, so it's already a very good indicative,

**possessed** · 10-13-2011, 02:18 AM

how about l3 latency result, do they improve using 4cu/4t or they still look slow?

**bondhahnmrt85** · 10-13-2011, 02:37 AM

Maybe need to patch for running core priority.

Now : Core 0 --> Core 1 --> Core 2 --> Core 3 --> Core 4 --> Core 5 --> Core 6 --> Core 7
Right priority : Core 0 --> Core 2 --> Core 4 --> Core 6 --> Core 1 --> Core 3 --> Core 5 --> Core 7

CMIIW

Thread: AMD FX "Bulldozer" Review - (4) !exclusive! Excuse for 1-Threaded Perf.

Thread Tools

Search Thread

Rate This Thread

Display

Tags for this Thread

Bookmarks

Bookmarks

Posting Permissions