VPCZ1 and cleaning Quad-SSD. is it possible?

After using my VPCZ13 I noticed that (most likely) due to lack of processing TRIM command, SSD performance was degraded. Initially it was almost 1GB/s and (using same test s/w), now I get (after fresh win7 reinstallation) below 500MB/s. That votes for cleaning SSD. I made a BIOS hack long time ago, so I tried to follow known procedures : broke RAID, changed RAID to IDE in my hacked BIOS and tried everything I could - HDDerase (all version 2.7 to 4.0), SUTIL (Samsung utility - since this quad-SSD is built from four Samsung drives) and few others. Nothing works. SUTIL says (when attemtping to low-level format) "Error: Busy time-out". All other software just shows nothing. After leaving HDDerase (with this nothing), eventually CPU gets overheated (fan is at crazy speed, and it blows outstanding heat from the side). Is there anything that will work on VPCZ1?

Thanks for all help.

Vojtek

 

No z1 ever reached speeds close to 1gb/s. 500mb/s is fine for a z1 quad raid. All those tools you mentioned might break your ssds rather than helping you and changing the ssd values in the hacked bios may render your z1 unbootable so you have to temporarily disconnect the ssds in order to get into bios again.

I have a z1 almost 3 years and the ssds garbage collection does a good job. There is no real long term degradation for me. I can prove that TRIM works on the z1 too (with another internal bios hack I made and 11.x and up Intel RST drivers) but it does not give any significant improvement over the included garbage collection at all. Trust me or not. Leave your ssds alone.

 

I have to disappoint You. First of all - I don't have any special software to test the speed. I just use TrueCrypt. and - at it's setup - I can benchmark (test) the system drive to check what encryption it will support at what speed. so I did. right after fresh windows 7 installation (and all drivers installed). I had (at best) 400-500MB/s (AES). this was before running secure clean on SSD (how I did that - read below). I double-checked that with WAT (it showed 7.0 for my system disk).

After running secure clean on SSDs, I reinstalled windows again. and from TrueCrypt, for same AES encryption I get 1.4GB/s. WAT now says that my system disk is 7.9 (top score). also - from user point - everything runs much faster. and it's something You can really feel.

What I did to secure erase? all recommended HDDerase and SUTIL are worthless - s/w written before era of SATA You can skip at once. the only thing that worked is parted magic. works like a charm. secure erase on SSD drive took 9 seconds per unit (since the command is built-in to SSD firmware). from my experience - I will always use it for win7 reinstallation.

Vojtek

 

I have to disappoint you, sir. The benchmark function in TrueCrypt is running encryption/decryption tests only on the cpu, storage drives are not touched by that. 1.4 gb/s is normal for AES on an AES-NI enabled Arrandale. If you select Twofish, Serpent or whatever that's not accelerated, you get much less.
You get those results even on computers with HDDs that max out at under 100mb/s.

I don't know what happened so you got crappy speeds before, but you were probably running on low clock or had your cpu busy when running the benchmark.

1.4gb/s are ideal values for the CPU, given that it fully load all 4 threads with AES and nothing else is taking cpu. You can test it by clocking down your cpu (throttlestop) or running cpu intensive tasks while benchmarking. You'll get lower results.

About truecrypt: If you encrypt your system drive with truecrypt, your z1's max seq speeds will drop down to around 300mb/s. Unencrypted, max seq speeds of z1 Samsung or Toshiba arrays are approx 500 (quad) and 350 mb/s (dual).

Read speeds have nothing to do with write speed degradation, trim, garbage collection, secure erase etc pp.

WEI is inaccurate BS. It's a dynamic benchmark and the grades are updated over time so your numbers should decrease over the years. If you fresh install win7 and run the test, you might get 7.9 on storage, if you connect to the internet, run all windows updates and run the test again you might get 7.3 only (it's still the same speed actually). That is simply because there are newer gen3 ssds that outperform your z1's array, and those ssds get the higher numbers now.

Believe me or not, I have been contributing quite some Vaio Z stuff here so I know what I am talking about. The z1 is still very fast compared to the average and in the end, random reads/writes are much more determing snappiness and speed of a system.

 

The SSD performance my 3 years old Vaio Z1 is still very good
The internal GC seems to work very well without TRIM

Normally, GC needs TRIM to notify which pages are not longer needed by the OS

in Vaio Z, I never know how GC works without Trim, my best guess is:
There are probably 300G true SSD storage in Z1's Quad SSD; but you are only using 256G, leaving 44G free/hidden
Whenever you write a new page, the SSD firmware allocates a new block from the 44G free/hidden space and the page is written to this new block at full speed
then the GC will copy existing allocated pages from the 256G visible section to this new block in backgroud
once a whole allocated block from the 256G section is copied over to the new block in the 44G section
the allocated block in the 256G section can be freed up and move to the 44G hidden section

 

Are you not comparing "apples and oranges?" (see dictionary of American idioms ) your 1.4GB/s from TrueCrypt is measuring cpu speed, not SSD. Why not use conventional SSD speed tests (Crystal Disk Mark, ATTO, etc) and compare your SSD speeds from these benchmarks? I don't think Z1 quad SSDs ever got much more than 500GB/s, brand new out of the box. I have a 3 year old Z13, dual SSD (256GB) and it got about 375 GB/s new, now gets 350/s and the drive has about 30% capacity still free. I wouldn't take to much stock in WEI (which I assume is what you mean by WAT); it's highly unreliable and it changes with newer SATA III drives. Finally, I don't think you should be all that concerned with your sequential write speeds unless you mostly transfer very large files at a time. The speed variable that produces noticeable SSD variations are the small random reads, which are excellent at 20GB/s.

Also, there is a rule of thumb on the NBR Sony Forum: pyr0's word is law, lol.

EDIT: I wrote this response a day ago and never hit send. I see that pyr0 has preceded me with the same response. I defer entirely to him, though I swear I didn't look at subsequent posts when I sent this a day later! He taught me everything I know - and then some.