How To Improve SSD performance on Intel Series 4, 5, 965 Chipsets (JJB Tweak)

answered by pm.

 

I did some more research and it turns out that I'm getting good benchmarks even without JJB tweak & 'Turn off Windows write-cache buffer..' checked when the CPU is under light load (to reproduce this I ran Prime95 with 4 threads).. So, sadly enough but it appears that my recent results which showed some boost are not quite accurate because I had some other processes running in background which could cause that improvement. Having said that, I applied the JJ's changes again and made sure the CPU load is 0-1% (no any background work) and re-ran the CDM (with the Windows write-cache buffer checked) - same low results as before. Will see what I get after the full reinstall / quick formatting.

 

Thanks for the info. But your drive is 2x bigger hence it has better metrics.

 

Hello I have a newly installed Intel 310 80GB SSD in the Lenovo listed below.

Here are my benches:


Normal Boot:

By meetloaf13

In SAFE MODE:

By meetloaf13

My Processor is in Family 6.

Would these fixes apply? My write speeds for the lower three benches seem low (almost 50%) of where they should be.

See:
Intel 310 Series 80GB SSD Review – Benchmark Testing | The SSD Review

Thanks for any suggestions!

 

Your results seem fine to me but you can certainly try out the JJ tweak. They are using a desktop machine to take the benchmarks which means it will always perform better than in a laptop.

Plus, you consume more than a half of the disk space (67%) which also negatively impacts the result.

 

Here is a clean install with no tweaks at all. I would give anything for this laptop to be SATA 2.0! I wish I would have installed this SSD in my son's Lenovo T410 and installed W7. I would love to see how this drive runs then.

My user speed is awesome though. Everything opens in a blink of an eye, start up are 17 seconds, so I guess I can't complain.

 

First of all I would like to say that I stumbled across this post by complete random.

But Holy $#@$!!!

Here is my results:

Before tweaks (Balanced profile)


After the power plan tweaks from page1



Now I just need to double check nothing is messed up with my turbo boost!

Btw I have an i7 620m (duo core)

EDIT: I just realisted this tweak never will mess with TB and it all seems to work : )

 

@ Thomas, the 50MB setting can lead to inflated results when the files get cached. Better test with 3x 100M in my opinion.

 

Well, the files would have been cached to benefit the non tweaked.

I ran the tweaked test first.

 

I get your point and I'm still saying test size 50 MB leads to unreliable results in my experience. But do as you please of course

 

Just to report back, this tweak did absolutely nothing for me on my Family 6 i7 740QM.

Thanks for the thread though.

 

Yeah I know what you mean mate heh for now as long as it doesn't make anything worse.. im fine with it. I wish there a way to actually crank more power out of this cpu... even if it means disabling hyperthreading so the turbo boost could go higher and what not.. (none of that makes a difference btw)

 

Sorry folks. I am confused.

The initial 6 regedit changes are trivial but the rest is confusing.

Edit: The Advanced Power Plan Settings and Processor Power Management settings are not accessed from within the registry nor from the battery icon (at least on Lenovo ThinkVantage Power equipped laptops), but from Power Options in Control Panel.

Second Edit: Thank you for all your work. SSD Benchmarks are up about 20-25%!

 

I just got my SSD and I was getting low 4k reads so I ran a few HD videos and ran As ssd at the same time, My reads jumped from 14.5/23 to 19.4/39

Since the CPU usage was around 70% this tweak shoudl work quite well for me then? ( I havent tried the tweak yet, just ran the cpu usage up with the videos)

 

I ran CrystalDisk on my Intel SSD (this is before the tweak), my read speed was about the same as JJB got before his tweak as well, however my write speeds were significantly slower. For example on Sew I got 214.8 read, and 84.55 write (as aposed to 107.9 write on JJBs test). For 512K I also only got 66.82 write, 4K 21.73 write, 4k QD32 67.88 write.

Is this normal? Is my SSD already losing life or something?

EDIT: Don't pay any attention to my message, I didn't realize people with all sorts of SSDs are trying this tweak out, I thought it was just the intel x25m 160GB on envy 14. Again, sorry

 

Jesus, sorry people theres something wrong with me today. I quoted the wrong message. sorry

 

While I did check my temps before and after the tweak (they were about 1-2c higher on cpu and TZ00/TZ02), it seems the bottom right corner on my envy is quite a lot warmer then before which would indicate more heat coming from the SSD. It doesn't bother me much but I'm just wondering if this is going to decrease the life on the SSD at all?

 

Questions:

Do these tweaks work for mSATA or just SATA?

Does mSATA use Intel SATA drivers or Windows default?

 

It's clear that those tweaks produce an increase in numbers in terms of benchmarking, but would I feel any improvement in terms of daily use???

 

That's exaclty what I would like to know as well. If I buy one SSD to upgrade in my HP HM55 chipset notebook for normal use, daily use, do I need to apply this tweak or this is just in case for benchmarking?

For normal use has a problem of performance or any other kind of problem? And does anyone knows what is the best recommended model to HM55 chipset?


Thanks.

 

You've got to test it by yourself to know for sure.

In my case:

Windows boots like 3 ~ 5 seconds faster (16 ~ 20 seconds instead of 20 ~ 25) (reads and writes);
I reinstall a small program on a regular basis, installs in 7 seconds instead of 12 (writes);
My antivirus full scan my drives in ± 12 minutes instead of 15 (reads)...

 

I think I will post some results of these tweaks:
Before:

After:


Big improvement of 4k read speed,16 to 22!!Not much change of other values, some even decreased a little, but that's within the margin of errors.

 

Eh it worked okay for me i guess


LOL

 

LOL nice RAM-PACK Dude !

 

haha lolol

 

Your results are a little low though; see mine...

 

DDR2 Bro lol
but for real i did this with a mechanical drive and it stays a strong 60 mb/s i know thats terrible but for a 5400rpm drive not really

 

Thanks! Disabling C3/C6 states was great too but only TurboBoost I had was +1 on all cores. Using your registry I had same results as Anseio in his 1-st post but setting to 100% instead of 95 gave me 21+ reads which is almost same as what I had with stamatisx's help.

EDIT/// NO! It blocks my TurboBoost the same!

EDIT: Found The REASON! If you want to have TurboBoost enabled you should set Proc. idle promote threshold less than Proc. performance core parking overutilization Threshold at least on 1%. But more is better.

Therefore I actually don't understand why to tweak Proc. idle promote threshold at all. If I understand coorect then when processor works at lower than this 95% you set then CPU starts to idle even more! Shouldn't it be lower than default 60 to have better results? I am waiting for an any answer guys!

 

Hey guys! Any idea what to do with 6 series chipsets? I'm getting 20/38 4k speeds on a Crucial M4, which should probably be 50% higher.

Thanks in advance.

 

Me too.

Same conditions.

 

Could anybody help me restore to default values before the tweak ?
I did the tweak about 4 months ago, but I think I've lost my original registry backup.
Anyone remembers what are the default values for each setting in the power plan ?

Thanks

 

If it's some standard / built-in power plan (e.g. Balanced, High Performance, etc) then you can click Restore default settings for that plan in the plan settings.

 

JJB tweaks actually seem to have slowed things down on my Qosmio X775

Follow up, the JJB tweaks do help a bit after all.

What really woke up some of my benchmark numbers was to turn off superfetch. Over at www.tomshardware.com there is an article that shows how to make the registry edit..

My 4K QD32 numbers are still in the 20's though. Any idea whats up with that?

 

Ive scanned through a lot of this thread but cant find any benchmarks regarding battery life.

On the new power scheme you set after making the processor settings visible, have you noticed significant percent decrease in battery life?

 

Yes (too much for me to keep this tweak for more than testing purposes.

 

You fix this by just setting those tweaks for the PLUGGUED TO A/C Adapter option only; you leave it as it was for the Battery operation...

 

had a play with all these tweeks with mixed results

my throttling/turbo is still working spot on under the powerplan i ear marked for the tweeks which is a good thing!

my ATTO results became way more erattic......but faster

AS SSD results showed a slowdown in all areas bar the 4k which went up as per the tweeks promise.

so the two benchmarks dont align results at all!lol!

 

Hey guys,

I have a Clevo W860CU with an i7-Q820QM, GTX260M and a Crucial M4 128Gb 0009.

Just tried JJB Tweak and i have to say that is GREAT!

Look at this tests, just made them switching the power modes from the custom I created to the normal high performance I was using before.

 

Does this apply to the Intel 6 series chipsets also? Like my Sager NP8130.

 

It seems that on my NP8130, doing this mod resulted in a very strange effect. My F@H time for a work unit went down from 4hrs 32min on GPU to 4hrs 3min. Yes that's not much, but that shows that when doing VERY heavy computations that opening up the CPU a bit is instrumental in getting a healthy speedup in the GPU computing.

I also noticed my Vantage Score went from P9301 to P9419 (a 1% boost, but hey... who's counting? )

Jason

 

What was exactly "strange" about that?

 

I guess the "strange" was that I wasn't expecting about a 15% GPU boost in folding and a 100 pt boost in vantage just on CPU. It makes me wonder what's hiding in the GPU of many computers that can be tapped without OC'ing.

Jason

 

I finally got around to applying the tweaks. I'm limited to SATA 2. Clean, new install on new Intel 320, 120GB.

First one is before applying the tweak (but had done disabling Superfetch, etc.). On the right is with the tweak.

The 4k score improved, but not as much as I had hoped for the read.

I do have the HM/PM55 issue. Downloaded newest chip set drivers and installed it first when I got to the desktop.

Suggestions?

 

Thank you very much !

 

Thanks so much for all your hard work folks.

Just installed an OCZ Vertex III 240GB (2.15) and following these tweaks:



I'm very pleased with the results.

 

I am soon getting an intel 320 series 160gb ssd (ordered) and was just doing some extra research. I came a cross this mod and also one where I change intelPPM to 4 instead of 3 in the registry.

From what I understand the intelPPM one shouldn't affect me negativity as I am using a 920xm. is this the case? Also it seems that the intelPPM mod is not as effective as this mod. Is this the case? should I apply both, or just one if so what one?

I was also wondering if this mod (or the intelPPM mod) will effect the performance of my computer (not including the ssd), as in will it make it slower or faster at idle/load (ts bench hasn't changed at all from before this mod), what about gaming performance?

Thanks, Daniel.

edit: I did a few more tests and i found that I couldn't get to my maximum turbo boost multipliers using these settings (not due to throttling). My multis were capped at 24(3.2GHz) which is my current 4 core active multis. This sgests that windows was not parking cores even when they were not needed so my multipliers never got above the 4 core ones and could not reach the 27x multi that I have set for when one core is active (yes i did see this multiplier before making these changes). I changed the settings back and now turbo boost and core parking is working correctly. once i get my drive i will try changing all of the settings listed except the onece rating to core parking and see if i still get an increase in ssd performance. I will re edit this once I try the intelPPM mod without the ssd to see how it alters raw cpu performance.

edit2: After the intelPMM mod it seems force all cores to be active at all times. I understand why this would help a solid state drives performance but it also means that I can only get my 4 core turbo boost at a maximum. I dont like that. It was also recommended that I turn off speed step and c state control in throttlestop. When I did I got turbo forced across all cores at all times! No dynamic increasing of clocks, just sits on 24x multi at all times on all cores (confirmed with throttle stop core temp and hwinfo). I really wish intel thought of a proper way to do this but what is is, i will wait until i get my ssd (in transit) and then do some more tests. hope someone finds this useful.

 

Why are these mods not done one desktop computers with ssd's?

 

I just thought of something you guys may want to try. I remember a while back for the hell of it I enabled on the "fly fsb detection" in coretemp. It put a constant light load on my cpu (it didn't effect performance at all or temps). This constant light load may be anough to do the same thing as the "iTunes mod". Maybe someone can give it a try. I will once mine arrives.

 

Hi all,

I just tried these tweaks and seem to have gained no difference at all. Running on 3820TG, I tried the disabling driver tweak on another thread and boosted from R:16-W:25 to R:26-W:45, but this tweak ended up with the old scores.

Note I created a power saver plan and included these settings to ensure minimal power loss, but made no difference?

Any ideas?

 

Got the High Perofrmance plan to work but when invoking the same criteria into the power saving plan seems to have no gain in SSD performance.

Is there any way to have the SSD performance of High Performance profile in the power saving mode?

General use in power saver mode seems to last me 5-7 hours use whist high performance is closer to 3-4 hours use.