• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

AMD hot fix just boosted my sandy in cinebench

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Benchmarks are getting old. A CPU is for productive work. not benchmarking and getting happy or stability testing..... Use your Rig and do your work and enjoy it intstead of looking at some numbers....... gl

You do realize this is AnndTech, right? Quit thread crapping.
-ViRGE
 
Last edited by a moderator:
Diceman2037 said:
safe mode is perfectly fine for performance testing a cpu, unlike older versions of windows, vista and 7 keep dma enabled for memory and disk access
Click to expand...

This is a very fine line....

If you have so many bloat ware in which safemode will actually give u a better score i think its time to reformat.

tweakboy said:
Benchmarks are getting old. A CPU is for productive work. not benchmarking and getting happy or stability testing..... Use your Rig and do your work and enjoy it intstead of looking at some numbers....... gl
Click to expand...

whatcho talking about willis?

My PC is art, a form of expression, and a way for me to be happy with a hobby, which is why i dumped so much money into everything about it.
Yes its a production machine... but it doesnt mean we have to conform the the apple way and everyone have the same generic system.
 
Guys its been a while with the first release AMD hot fix on my intel rig and its 100% stable and HT performance is up in every bench.

I even get higher gigaflops in intel burn test.
 
"benchmarks are getting old" lol, that is gold

anyhow.. wonder if this hotfix would help my e350 laptop at all? I dont have high hopes of it being able to play HD netflix, but you never know.
 
zCypher said:
"benchmarks are getting old" lol, that is gold

anyhow.. wonder if this hotfix would help my e350 laptop at all? I dont have high hopes of it being able to play HD netflix, but you never know.
Click to expand...

Afaik this kind of optimization only aids designs without totally dedicated 'cores' for each thread. As in Intel HT cpus and AMD 'module' designs.
 
I guess the new 2 part update that was just released is not compatible with intel systems?

Tried to run both with the popup"the update is not applicable to your computer".

Or is the original file still floating around?
 
skipsneeky2 said:
I guess the new 2 part update that was just released is not compatible with intel systems?

Tried to run both with the popup"the update is not applicable to your computer".

Or is the original file still floating around?
Click to expand...

The patch "works" for my Intel CPU (i7-2670QM). Don't bother using this on a non BD CPU.

Geekbench benchmark before patch - 8900~
Geekbench benchmark after patch - 7900~

Aida64 before patch (CPU benchmarks) -
CPU Queen 33953
Photoworxx 49096
CPU ZLib 191.4 MB/s
...And so on.

Aida64 after patch -
CPU Queen 29813
Photoworxx 46941
CPU ZLib 170.9 MB/s
I could go on, but I see no improvements and hope people get the point.
 
KayVeeEmm said:
The patch "works" for my Intel CPU (i7-2670QM). Don't bother using this on a non BD CPU.

Geekbench benchmark before patch - 8900~
Geekbench benchmark after patch - 7900~

Aida64 before patch (CPU benchmarks) -
CPU Queen 33953
Photoworxx 49096
CPU ZLib 191.4 MB/s
...And so on.

Aida64 after patch -
CPU Queen 29813
Photoworxx 46941
CPU ZLib 170.9 MB/s
I could go on, but I see no improvements and hope people get the point.
Click to expand...

why dont you try real benchmarks that support ht and use 8threads and see what the results are.the whole purpose of this patch is for 8+ threads.

Iv emailed this to many peeple and everyone has had there numbers go up while using 8 threads on a sandybridge cpu

I have a dual boot os right now and ill bet you 100 bucks that my numbers are up.Ask me a bench to run and ill post before and after numbers with the 2 different OS

just ran cinbench on my bloated os with the patch and everything running in the backround,full msconfig running everything and no os tweaks at all.virus scan,network junk,dvix,smart burn,all the junk I have on my normal windows install.

got 70 fps in open gl test and 9.21 in cpu bench @4.7ghz swapped over to my clean bench hd that is running nothing on boot and setup for max performance and got 66 fps and 9.15 in cpubench and I didnt change any overclocking settings,just went into bios and picked the other hd.The clean install is also on a way faster 550mb/sec force GT ssd vs a 280mb/sec ssd on the bloated install.I also ran it 3 times in a row and could not break 9.15 on the clean drive.

to get .6 more points in cinbench you would have to clock the clean install another 150-200mhz higher to gain that much from just a patch

cpu qween? that bench ran in under a second and Igot the same numbers on both systems 588XX litterally that bench finished in like half a second
 
Last edited:
grkM3 said:
why dont you try real benchmarks that support ht and use 8threads and see what the results are.the whole purpose of this patch is for 8+ threads.

Iv emailed this to many peeple and everyone has had there numbers go up while using 8 threads on a sandybridge cpu

I have a dual boot os right now and ill bet you 100 bucks that my numbers are up.Ask me a bench to run and ill post before and after numbers with the 2 different OS
Click to expand...

Waht the patch does it make sure the application maps its thread differently depending on the workload.

so you get improved 8thread behavior and a lot worse low thread behavior? that would be a explainable and unacceptable for every SB user imo.

Depeding on the workload threads might be sheduled per Mdoule or per core. If the same anology is applied on SB (or intel cpu's) you will effectively have cases where you choose to run 2threads on 1core (+HT) instead of 2 cores like it is now. Which would be a huge loss in performance for SB.
 
Riek said:
Waht the patch does it make sure the application maps its thread differently depending on the workload.

so you get improved 8thread behavior and a lot worse low thread behavior? that would be a explainable and unacceptable for every SB user imo.

Depeding on the workload threads might be sheduled per Mdoule or per core. If the same anology is applied on SB (or intel cpu's) you will effectively have cases where you choose to run 2threads on 1core (+HT) instead of 2 cores like it is now. Which would be a huge loss in performance for SB.
Click to expand...

no I get the same exact performance in 4threads also,this patch is an early gift that windows 8 will give all to threaded cpus on windows 7.

any special bench you would like me to try? Im on the clean os now and will run a few becnhes since Im sick and cant sleep lol
 
Last edited:
Arkaign said:
Afaik this kind of optimization only aids designs without totally dedicated 'cores' for each thread. As in Intel HT cpus and AMD 'module' designs.
Click to expand...

Depending on what areas of the scheduler have been targeted for improvements, this could improve performance in low-threadcount situations even with multi-core CMP microarchitectures.

Thread migration causes performance degredation, if the scheduler reduces thread migration then it will be a tide that raises all boats.

That said, I don't know what this specific set of patches does, and in reading this thread its not clear to me that anyone does.

Hopefully Anand does a nice review of the patch, as he did with the TLB patch and eventual stepping fix on Phenom.
 
grkM3 said:
intel burn test gflops is up bigtime with hyperthreading and avx enabled.I broke 112gflops at only 4.9ghz

I couldnt break 100gflops at that speed before the hotfix
Click to expand...


Nice! Just think, with a little more optimization you might beat the 120+gflops my 2500k gets @ 4.5ghz 😛
 
grkM3 said:
intel burn test gflops is up bigtime with hyperthreading and avx enabled.I broke 112gflops at only 4.9ghz

I couldnt break 100gflops at that speed before the hotfix
Click to expand...

Puppies04 said:
Nice! Just think, with a little more optimization you might beat the 120+gflops my 2500k gets @ 4.5ghz 😛
Click to expand...

It's a hyperthreading thing. For some reason it scores lower in the GFlops with it enabled. Guess HT has advantages and disadvantages 🙂

Here's a screenshot of my 2700k at 4.8ghz with HT disabled in bios. The temps and voltages are with prime running in the background to get vcore under load and to see the whimpy temps prime generated compared to IBT 🙂

48ghz_NoHT.png
 
Kenmitch said:
It's a hyperthreading thing. For some reason it scores lower in the GFlops with it enabled. Guess HT has advantages and disadvantages 🙂

Here's a screenshot of my 2700k at 4.8ghz with HT disabled in bios. The temps and voltages are with prime running in the background to get vcore under load and to see the whimpy temps prime generated compared to IBT 🙂

48ghz_NoHT.png
Click to expand...


Hehe I know, hence the 😛 in my post.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top