Interesting SSD performance find...

[stringfellow]

I had a major drop in performance for my Intel 25v SSD over the last few weeks.

I tried all of the usual performance tweaks/fixes/refreshes. Since the whole SSD phenomenon is relatively new, the same optimization advice comes up over and over.

After a couple of weeks of research combined with some dumb luck, I stumbled upon the reason: the SSD does not like it when I enable my C3 power saving function in the bios. That simple change dropped my performance by about 20 percent.

I tried to eliminate other variables--I increased my clock speed, bus speed, etc., etc., to compensate, but the only change that ever really mattered was the C3. So for any of you also with that problem, give it a shot and see if it works.

Along the way, I tried AHCI vs. IDE (based on the latest arguments for each respective setting) and didn't really notice a difference either way. However, I also tried replacing Microsoft's standard Win 7 AHCI driver with Intel's driver. At least for my Intel SSD, I got a small (5%?) bump in performance overall (i.e., when I disabled C3). But most of all, I regained some of the lost performance with my C3 setting. (my latest fascination is maxxing OC, while minimizing power so I want to keep the low power setting).

If any of you have any other advice on this, I would love to hear it, but hopefully my experience is helpful to you all.

 

[Shadow703793]

Good to know. Thanks for posting this.

 

[sminlal]

Yeah, that is interesting. Thanks for sharing from me too!

 

[Shadow703793]

@OP: Mind listing full specs. Could be related just to the motherboard/BIOS.

 

[elel]

Cool. I love threads which give interesting, original info.

 

[stringfellow]

Best answer selected by stringfellow.

 

[stringfellow]

Hi Everyone:

If this helps, here are my specs:

Intel i5 750 at 3.58 GHz @ .85v with Xigmatech 1283| Gigabyte P55A-UD3 F4 | MSI 260 GTX OC | 4 GB AData DDR3 1600 | Intel X25V | WD 6401 AALS Black | 600 Watt OCZ StealthExtreme | Antec 900 | Win 7/64

Everything is set to auto on my mobo. EIST is disabled. My bus speed is set at 149. Strangely, my mobo adjusts my multiplier to 24 (which I didn't think it could do, due to the i5 being locked at 20). The mobo has a feature that take the multiplier to 21 from the stock 20, but with C3, it will toggle between 21 through 24, depending on load. As I said, I am pretty pleased with being able to hit such a high overclock on such low voltage and heat (still gives me ~11.5 pi on super pi).

If there is any other info that I could provide that could help you to figure this out, please let me know.

One thing that I suspect, is PCI bus speed. I understand it is something that generally should not be monkeyed with, at risk of frying your GPU. My Mobo handles PCI speed automatically, and supposedly adjusts it. I wonder if it "downclocks" it in any way with C3 enabled. I look to those smarter than me for some answers...

Any other suggestions are welcomed...

 

[EXT64]

That is Turbo (21-24). However I find the 0.85V extremely hard to believe at full load. Use CPU-Z to watch the voltage when stressed (say, w/ prime95 or the like).

 

[stringfellow]

That is Turbo (21-24). However I find the 0.85V extremely hard to believe at full load. Use CPU-Z to watch the voltage when stressed (say, w/ prime95 or the like).

Hi EXT:

You are absolutely right--it cruises at about .82 for most cycles. If I push it, let alone with something like prime95, it goes to about 1.2 volts. But the key is the hottest core stays around 32 C, lowest around 27. I am pretty happy with that--for the most demanding of tasks I do (Battlefield Bad Co 2), my ~3.5 GHz is probably nowhere near bottlenecking my GPU.

I have found that if I go from a bus speed of 149 to 150, my voltage jumps to about 1.1 V or so, from that standpoint 149 is optimized. Similarly, if I go much over the low 150s, I no longer get the benefit of turbo boost (i.e., my multiplier stays fixed at 21).

Now if I could just find a way to speed my SSD back up again... I have a feeling that once they have been around longer, people will find ways to truly optimize the SSDs via firmware, etc. At minimum, I am waiting for my Marvel SATA3 chipset to get a driver update.