I still don't think this removes the delay, just, er, delays it. The RAID expansion method is preferred method whether you plan on dual booting to Linux or not.
I wonder how much money this thread will save Sony. They wanted me to RMA my machine, which is unnecessary.
-
what do you think about the performance numbers.
after I went from 2 to 4 array the sequential read was over 600. then I expanded the C drive to include the unallocated space from adding the additional 2 to the array, ran the test again and the 600 went to about 450 - 470. the rest of the test was the same. Think I did the updates (windows and sony) after the first test and before the second one
Not sure why that would be the case, on the other hand, not sure how consistent the test is, I used that one as a basis of comparison since it seems to be used a lot in this forum. -
Your system is performing normally.
I'm not a huge fan of CrystalMark because it tests a section of the HD and not the entire volume. I normally see swings of 100MB +/- on the sequential tests. Any tool you can run 5 times and get 5 vastly different results isn't good, IMO.
I still run Crystal from time to time, but really rely on HD Tune Pro for comparison benchmarks. I even purchased the full version. -
TofuTurkey Married a Champagne Mango
I'm thinking this suggests that Sony messed up RAID config via BIOS (since they can touch that), but Intel can still get it right via their app.
I'm wondering if this BIOS error wasn't caught during laptop building because it's quicker to copy the image onto just a single disk first and again the app is used to migrate, so Sony went this route.
Edit: I suppose a question now is: without touching the BIOS, if I do an erase of the SSDs, can I still copy the image back onto the 'RAID' drive without the blinking problem appearing? -
Dear all,
This is my first post on this forum ever. While I'm still thinking on which laptop to buy (although the Z is on nr. 1 on my list at the moment), I'd like to say a couple of words:
After having read this forum's topics for a couple of weeks now, regarding Boot times, clean installs, Optimus, crappy Sony vid drivers, DIY VidDocks, etc., I'd really like to give a
FAN-CRAPPING-TASTIC BIG APPLAUSE
for the incredibly good work from a couple of guys (gals?) here. Most notably: Nautis, Beaups, but most of all Zoinks.
Other people, please don't be offended if your name isn't on this short list, but I'd just like to pause for a bit to give some credit here, where it's due!
I'm not yet entirely convinced to buy the Z (crappy Sony vid drivers are mainly the factor against), but seeing how much work is done here by these people would definitely be the deciding factor to go for it and plunk down the money!
Thanks again, and know that your posts are very much appreciated, even when the readers don't actually reply most of the time.
And now, get those #&@$(*& Optimus drivers to work with the stamina button Zoinks! (* Just kidding*)
-
I also think the reason it did not make it to the restore disk build is because they would have to tell the user to go into the bios and set it up for 2 disks first. It could be that the restore disk wont touch that particular BIOS setting is because the restore disk cant know exactly what disks are currently populating the PC and their intended use. So this was Sony's way out, they thought no one would notice -
I will work on updating the hybrid drivers, just sans-Optimus.
-
I would like to report a significant improvement in the boot time of my VPCZ118GX/S (128 GB x 3). My most recent boot took 45.28 seconds from pressing the power button until getting a normal Windows 7 cursor and involved only 10 cursor blinks from the BIOS screen until the Windows start screen appeared. Previously, I was getting 55 second boot times and 17 cursor blinks at best. The only thing to which I could possibly attribute this improvement is Diskeeper with HyperFast. However, the improvement was not immediate. It took a couple of days.
-
Interesting. Not sure how free-space optimization helps, but I purchased the product for another system. I'll give it a go.
-
I have some more test results for you.
I just did a clean install (after using Paragon backup to take a full raw image of the SSD including out of partition areas) of Windows 7 Pro.
I started by clearing the RAID array from the RAID option ROM. Then installed to a 32gb partition on just one of the drives (I have 2x64Gb). Then after first boot I installed the RST 9.6 drivers and then expanded the volume to use the second disk giving me a 122gb drive but still with just my 32gb partition on it. I choose a stripe size of 16kb (controversial I know!) At this point from power button to logon screen was 16 seconds.
After making another backup with Paragon I then added drivers, office 2007, visual studio 2010. I now seem to have a fully working system with a power switch to logon screen time of 22 seconds. Visual studio takes 3 seconds for first load. I am very happy
I choose the 16kb stripe because a) the RST help guide suggests it is the default when using SSD's and b) because I am not interested in high sequential throughput scores in benchmark program, I want fast boot and fast visual studio load/open/compile/debug/edit cycles. I figured that a smaller stripe would increase the parallel use of the drives during the smaller 4kb page in operations used when a program is launching (.exe's and .dll's don't actually get 'loaded', they get mapped into virtual memory space and the pages are pulled in as the different parts are accessed for the first time). Seems to be working well for me. -
Have you tried the "Tony-TRIM" method (free space consolidation with Perfect Disk + AS Cleaner or FreeSpaceCleaner.exe with the FF option checked) on the VPCZ1? -
I have. No difference that could be measured
-
So, after get close to 30 cursor blinks (Without ever doing a system factory reset) I decided to do one with the Discs I made.
After the factory reset had 24, than 17 blinks, than 16, now its around 12. Its slowly getting less.
Im happy for the results though. -
Update 2: This time, instead of simply updating everything in Windows Update, I noticed that Windows itself had unchecked 2 times so I didnt install them:
-Microsoft .net framework 4
-Update for win 7 x64 (kb971033)
Maybe those could have an effect....not sure -
Just thought I'd add another data point to the discussion.
-Peter -
Even after installing all my apps. -
i registered for this forum just to post on this thread. first and foremost, thanks to all for the hard work
i'm also having the same problems with boot on my signature collection (512gb, etc). sometimes can get up to 25-30 blinks, which, needless to say, is totally unacceptable for this kind of machine
question is: is there any consensus here on what to do that might help that doesnt involve fooling around with the RAID configuration? also, I definitely don't want to have to do a restore (plenty of programs and data here I don't feel like copying back
do you think it's worth bringing it into sony style and letting them try to do anything?
thanks for your help everyone -
-
No Windows settings or patches can do anything with this, because it happens before the OS starts to load (and thus also happens for Linux users). -
Now down to 11 click. Before I had 24-30 blinks. I have the quad setup.
I also have the suspecian that having a Quad raid is faster in read/write, but takes a bit longer to start up.
I think the normal performance should be 10-12 clicks for the Quad. Dual an triple are anywhere from 5-8 blinks. -
I did mention it to Sony's "Second Level" Technical Support. Naturally they dismissed it as anything I should be concerned about. -
-
Sony may have figured out a faster way to access the RAID markers and assemble the volume before boot. -
-
i have a quad set up with samsung SSD. i get like 6 blinks from a cold boot and coming out from hiberation. sometimes 7. so i dont think it depends whether its a dual or quad setup.
i havent done a clean install yet and i am using the machine for file sharing too. -
what ever happened to the idea of re-writing the RAID markers and clearing some space at the end of the drive
-
Well I kinda thing Sony should hve thought bout all of this, such a expensive machine and having to mess with raid, partition and all that stuff. Im happy mine doing great now though
-
-
I know it's unnecessary and borderline BAD to defragment an SSD, but has anyone tried defragmenting to see if that has any effect on the number of cursor blinks you get?
-
-
So what you do inside the file system, whether it's drivers or defragmentation won't have any effect.
Why people see diminishing blink times after booting a few times is likely for two reasons:
1: (Applies to SSDs only) Because the drives themselves have a small internal journal that they replay when you power them up (much like a journaling file system, but at block/sector level). To reduce the number of expensive erase operations, during normal operations, the drive controller will write new data to a temporary area and reply "OK, written" to the driver, even though the data hasn't yet made it to the final place on the drive.
When the "journal" is played back, any data still in temporary work areas will be written to disk.
With a fresh Windows install, there's a lot of write activity going on at shutdown. As you use the OS for a while, and reboot a few times, the OS has less need to update anything, which means less writes. Which means less to replay and commit when you power it back up.
2: (Applies to any kind of RAID drives) If the drives haven't been shut down cleanly with the RAID actually being taken offline before the final flush of the drives, the RAID will continue to be marked as "dirty", which means the controller will have to check the integrity of the RAID before assembling it. The fewer changes that have been made to the drive, the less time this takes.
But normally, the two above are minor operations compared to scanning for RAID markers.
The underlying problem is that RAID wasn't designed for desktop use.
On systems with a large number of drives, it's not uncommon for the BIOS and controller firmware to take several minutes assembling RAIDs before being ready to boot. An IBM server I use takes around 15 minutes from power-on until boot, and this is considered normal.
RAID was designed for systems that you rarely reboot, but keep running. And RAID 0, even though it isn't really RAID, shares this design, and relies on RAID markers for assembly, just like real RAIDs.
What's needed is for someone like Intel to come up with a laptop version of RAID 0, where the configuration is stored in the BIOS setup, and not on the disks themselves, and where the array is always considered "clean". That would allow instant assembly, at the expense of higher level RAID auto-rebuild. -
I've rebooted dozens of times over the past month, and I'm constantly at around 26 blinks. I think what you're saying is that there's no easy fix until Intel comes out with a change for the RAID setup on this computer. Is that correct? Other than that, i guess I'll just deal with the 26 blinks. It's still a fast computer; just not as fast on startup as you'd think it would be with SSDs.
-
Well, there is one easy fix, but it doesn't help much: Never let the machine power off from standby, and never let it hibernate. If it does, the RAID will always be "dirty", and extra time is spent to verify basic RAID integrity. So shut down all the way.
No, this won't solve the problem, but in my experience from desktop systems, it reduces the pre-boot time just a tad. -
I don't get this, I was all the way up to 30 blinks, now I'm in the 8-9 blinks.
No clean install, didn't remove anything but Norton.
But I did try to keep it as clean as possible without installing junk.
Weird... -
I noticed that you can even push the blink count down a little bit by setting the internal (solid state) disk as the device with the highest boot priority in the BIOS. With that, I get 6-7 blinks with my i7, quad-ssd setup.
-
-
Remember you have to change the order when you want to boot off the optical or external drives. -
Not...
Down from 27 blinks to 25 right now. Oh well, thanks for the assistance -
-
TofuTurkey Married a Champagne Mango
Perhaps we need a Plan of Attack
Let's say the problem is with the way the bits, which define the layout of the RAID structure, is located on the drives. And also perhaps what those RAID-bits actually are.
Assuming that, we might begin with a comparison of various laptops' RAID-bits. To do this, we need:
1. A way to capture those bits. The only way I know which might work is to do a dd via Linux of the disks. The bits captured have to be constrained to the start and end of the disks otherwise we'll run into privacy issues.
2. A sizeable number of samples. Having multiple users execute (1) and post their results.
Given (2), we can look at the bits for a well-behaving RAID drive, and compare that with those of a misbehaving one. It will also mean that we have a chance to copy good bits over bad ones, though that will probably require some testing + trial and error to see if that'll fix the blinking issue. -
if you have enough free disk space, what if you went into RST or RAID BIOS or whatnot, removed one of the drives from the array, let the array reconstruct, and then added it back?
could this fix the slowdown without the need for a total system wipe? -
If you remove one drive, you lose all data on all drives. -
TofuTurkey Married a Champagne Mango
EDIT: I don't know if it's possible to configure RST to move data off of a physical disk onto the rest, someone with Sony SSDs can fire up RST, go into that and check... -
Sony VPC-Z11 - Slow POST
Discussion in 'VAIO / Sony' started by ZoinksS2k, May 17, 2010.