AMD hot fix just boosted my sandy in cinebench

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
For a simple software patch with no negatives, sure it is.

I would can that patch "meh"..okay at the most.
think about it.
If this where a new CPU driver...that changed FPS from 9.3 FPs...to 9.5 FPS...I would respond in the same way.

Awesome has a whole different meaning:
http://dictionary.reference.com/browse/awesome

It's not impressive...it's not even the 10% people hyped tha patch up to be...so again:

lulz
 
Last edited:

grkM3

Golden Member
Jul 29, 2011
1,407
0
0
So an "improvement" from ~9.3ish to 9.47 is "awesome"?


lulz

yeah it is awesome since you have no clue what it takes to go from 9.3 to 9.47 in cinebench.

that gain I got is equivalent to a 250-300 mhz overclock.

so this awesome patch just made my 2600k act like its overclocked 300mhz.

LUZ omg lollll....clueless post and please next time try using cinbench and see what it takes to raise your score .17 points
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
yeah it is awesome since you have no clue what it takes to go from 9.3 to 9.47 in cinebench.

that gain I got is equivalent to a 250-300 mhz overclock.

so this awesome patch just made my 2600k act like its overclocked 300mhz.

LUZ omg lollll....clueless post and please next time try using cinbench and see what it takes to raise your score .17 points

So now a 2-300 Mhz OC is "awesome"...

The only place the word "awesomw" fits to this patch would be in AMD PR material...
 

grkM3

Golden Member
Jul 29, 2011
1,407
0
0
So now a 2-300 Mhz OC is "awesome"...

The only place the word "awesomw" fits to this patch would be in AMD PR material...

are you kidding me?

what does my results have anything to do with AMD?

I am running there patch on a sandybridge and in heavy threaded tasks it gained what about a 5.0ghz 2600k gets running at 4.7

I broke 10.4 at 5.3ghz with this patch and that is a pretty big gain with a 4 core cpu.

with this same patch zip 7 bench wipes the floor with a BD and that was one of the only benches that AMD had over intel and with the OS running the threads better to the cpu it boosts intel chips alot more than it does with BDs

BD saw like 3% tops and the gain I got in cinbench is like alomst adding another core.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
So an "improvement" from ~9.3ish to 9.47 is "awesome"?


lulz

I would can that patch "meh"..okay at the most.
think about it.
If this where a new CPU driver...that changed FPS from 9.3 FPs...to 9.5 FPS...I would respond in the same way.

Awesome has a whole different meaning:
http://dictionary.reference.com/browse/awesome

It's not impressive...it's not even the 10% people hyped tha patch up to be...so again:

lulz

So now a 2-300 Mhz OC is "awesome"...

The only place the word "awesomw" fits to this patch would be in AMD PR material...

Lonbjerg, did you come in here to do anything but thread-crap and harass the OP? :confused:

We have a "solution" for member's who fancy themselves as being an authority on what is awesome and what is not...:hmm:
 

grimpr

Golden Member
Aug 21, 2007
1,095
7
81
He clearly showcased his anti-amd bias without giving a damn about the thread or grkM3's personal and proved experience.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
As some of you know in some cases with hyperthreading you actually lose performance or when the program is threaded for it you can gain some performance so I decided to give the hot fix a try on my 2600k

I ran cinebench 2 times and got 9.51 and 9.52 and then installed hot fix and rebooted and ran it once and got 9.68 and then a blue screen on the second run.

Im doing some more testing now and it seems stable so far as my rig was setup to pull the least amount of watts so thats prolly why I got the blue screen at 5ghz

All the hotfix does is make windows less twitchy. Aka, it will keep a thread on a core longer without constantly shuffling it between cores to balance temperature. IIRC it also takes into account full core vs split core (HT or AMD's double cores).

The moment I saw it I said "this should improve performance for intel CPUs too!". And by I said I mean typed, in several threads on anandtech. I really wanted to see anandtech test the performance improvement of SB for it compared to BD; and posited that SB would get nearly the same percent boost from it.

The blue screen is becuase your OC is too agressive and should not have occured.
 

Diceman2037

Member
Dec 19, 2011
54
0
66
All the hotfix does is make windows less twitchy. Aka, it will keep a thread on a core longer without constantly shuffling it between cores to balance temperature. IIRC it also takes into account full core vs split core (HT or AMD's double cores).
no, thats not all it does at all, but the full details on the patch are pretty much non existant.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
no, thats not all it does at all, but the full details on the patch are pretty much non existant.

http://www.anandtech.com/show/5448/the-bulldozer-scheduling-patch-tested

The first update simply tells Windows 7 to schedule all threads on empty modules first, then on shared cores. The second hotfix increases Windows 7's core parking latency if there are threads that need scheduling.

According to anandtech the 2 BD patches do what I said.
Notice that:
1. That behavior improves things for BD for most, but not all scenarios. (it will not schedule 2 threads that share a resource on the a pair of shared cores).
2. That behavior IS the ideal for intel CPUs with HT for all scenarios and improves the condition compared to previous intel CPUs.
3. Whoever has inferior switching speed would get a greater benefit, and that is PROBABLY AMD. But nobody did a comparison of intel vs AMD benefit from this patch.
 

Diceman2037

Member
Dec 19, 2011
54
0
66
reducing twitchiness doesn't account for why slightly more performance is obtained when all the cores are peaked out.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
reducing twitchiness doesn't account for why slightly more performance is obtained when all the cores are peaked out.

sure it does.

Twitchy:
Thread 1 is taking 100% of core A
Thread 2 is taking 100% of core B
twitch
Thread 1 is paused to be moved to core B
Thread 2 is paused to be moved to core A
Thread 1 is taking 100% of core B
Thread 2 is taking 100% of core A

Repeat many times a second.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
except threads are pinned to cores in the case of cinebench ;)

the vast majority of programs do not pin threads to cores though.

However, indeed that would not explain the increase in cinebench performance. What about the other patch? prioritizing empty modules over shared cores/HT.
 

Diceman2037

Member
Dec 19, 2011
54
0
66
the vast majority of programs do not pin threads to cores though.

However, indeed that would not explain the increase in cinebench performance. What about the other patch? prioritizing empty modules over shared cores/HT.

never installed the core parking patch, only the leaked and rerelease of the first patch.
 

Diceman2037

Member
Dec 19, 2011
54
0
66
Windows has gotten another update that is installed via automatic updates, KB2679255. This updates Ntkrnlpa.exe and Ntoskrnl.exe to 6.1.7601.17790(GDR) (or 6.1.7601.21936(LDR))

Anyone tested yet to see if the performance improvements from KB2645594, 6.1.7601.17730(GDR)/6.1.7601.21866(LDR) have carried over?
 
Last edited: