Thinkpads Forum

erik wrote:my samsung SSD in the X6 ultrabase tops out at 74 MB/sec using an OEM lenovo SATA HDD adapter.   native it hits around 105 MB/sec.   i don't understand why your ultrabay is so slow. :??:

i'm anxious to see your W700's RAID 0 array.   when do you get the machine back?

My T61p has the "ThinkPad Serial ATA Hard Drive Bay Adapter" 40Y8725.
Perhaps Device Manager is giving us a clue, see here:
http://picasaweb.google.com/tinkererguy ... 6339021794
Note the internal C: drive is showing without ATA or SCSI or anything, but the Ultrabay mounted D: is showing as ATA. So there is some difference in the attachment to the system, but I still can't explain why such a dramatic slowdown. It's perhaps treating the Ultrabay device as a USB 2.0 attached device or something? Note sure.

As far as the W700, it'll be a while, see my related post here:
http://forum.thinkpads.com/viewtopic.php?t=68840

Little experiment, read the posting images faq here:
http://forum.thinkpads.com/faq.php#24
trying the embed image option from Google Web Albums, not sure if this forum allows that. Tried pasting content from "Embed image" area of photo information area of Google Web Album, then clicked the "Code" button here when drafting this post, then pasting, but it's not working. Anybody know the right way, or if embedding can be done at all on this forum?

not sure but maybe you need to enable HTML code in your postings?
(if I remember correctly there is some option for this in your profile for this forum.)

Thanks for the tip, (html was disabled in my profile, turned it back on).

So, let me try this again:

Here's the "Code" button method: Here's the "Img" button method, by right-clicking the actual web album photo, then pasting in the URL
http://lh4.ggpht.com/_vroM1SelIJI/SQkKL ... Drives.jpg
instead of embedded code, here's the result:


That didn't work either, that is, red x shows instead of an image.

Also tried the "Img" button using the link the Picasa photo properties site provides:


No good either. Oh well, seemed worth a try.

tinkererguy wrote:Little experiment, read the posting images faq here:
http://forum.thinkpads.com/faq.php#24
trying the embed image option from Google Web Albums, not sure if this forum allows that.

You're looking at the wrong FAQ. That's the FAQ for the phpBB software. The two that count are in my sig line, and the rules one has some very specific instructions about (not) posting images.

nonny wrote:
You're looking at the wrong FAQ. That's the FAQ for the phpBB software. The two that count are in my sig line, and the rules one has some very specific instructions about (not) posting images.

OK, no problem, I won't try to display or embed photos at all anymore.

I noticed btw that the URL for checking ezserv status has changed, here's the new one:
http://www-03.ibm.com/support/ezserv/home.jsp

Have now also read your entire Rules of the Road:
http://forum.thinkpads.com/viewtopic.php?t=14339

Found this relevant sentence:
5. IF you must add a photograph to your post, PLEASE limit the size to 50k or less and place a warning in the subject line.

I'm sorry I didn't follow this rule: at least the photo embedding never worked anyway.

For what it is worth, my idea of using the RAID 1 configuration to upgrade my drives is working out just fine.

I ordered my W700 with RAID 1 and a pair of 5,400 rpm 320GB drives. Lenovo delivered my machine with Fuji drives Lenovo P/N 42T1090.

I got a pair of Seagate ST932041AS drives from Zipzoomfly for about $185 for the pair.

After part way configuring my W700 with the Fuji drives, I removed one of the drives (in bay 1), swapped the caddy and rubber rails over to a Seagate drive and rebooted.

Then I had to go to the RAID Array BIOS and tell it to use the new drive in the RAID 1 Array. Then I booted to Windows Vista and the Intel software took over rebuilding the drive. It took about 3 hours.

Then I removed the remaining Fuji drive from bay 0, slipped the first Seagate drive that the RAID 1 configuration had just rebuilt into bay 0, put the caddy and rubber rails on the second Seagate drive, and rebooted. Again I had to go into the RAID array bios and tell it to use the new blank drive in the RAID array, then boot to to Windows. Now, two hours into the process, the second drive is 75% rebuilt.

This is really painless. I am happy. I think I will buy another pair of drive caddies and rubber rails and keep the 5,400 rpm drives to build backups periodically to keep off machine. This is faster and better as far as I can see than using GHOST or Acronis TrueImage.

Contrary to the W700 hardware maintenance manual, the two drives in the RAID 1 array do not need to be the same FRU number. My drives rebuilt even though the rpms were different and the manufacturers were different.

The Seagate drives have a slightly different connector than the Fuji and Hitachi drives I've used in the past. There is a third section of pins. These are missing on the Fuji and the Hitachi. I peeked inside the drive bays and saw that there is nothing in the bay to match up with these pins and nothing that physically conflicts with them.

I took a couple pictures and I'll try to post them later on my google site with a link.

Ok, great, glad the procedure worked!

Cloning drive to drive in Acronis probably would have been a bit faster, but I like the simplicity of your approach.

Now, drumroll please, eager to see what kind of speed you get with "ATTO Disk Benchmark" (see previous page of same thread for download details).

As a reminder, results from my T61p for comparison:
http://picasaweb.google.com/tinkererguy ... 0405449714

Rebuilding from the Fuji 5,400 rpm to the Seagate 7,200 rpm took 3 hours. Rebuilding from one Seagate 7,200 rpm to a second Seagate 7,200 took about 2 hours.

I will run that benchmark when I'm able to focus on it later this evening.

barrywohl wrote:...
The Seagate drives have a slightly different connector than the Fuji and Hitachi drives I've used in the past. There is a third section of pins. These are missing on the Fuji and the Hitachi. I peeked inside the drive bays and saw that there is nothing in the bay to match up with these pins and nothing that physically conflicts with them.

I think this third set of pins is used for jupper settings. See also the graphic here.

Here's more info on the jumper settings.

basketb wrote:I think this third set of pins is used for jumper settings. See also the graphic here.

Here's more info on the jumper settings.

Exactly correct.

By the way, my "Windows Experience Disk Data Transfer Rate" score went up from 5.3 to 5.8. My overall windows experience went from 5.3 to 5.8 also as all the other tests are at 5.9 already.

"barrywohl"
When you have a chance, could you post your complete Windows Experience Index results at the existing thread for this, over at:
http://forum.thinkpads.com/viewtopic.ph ... 575#463575

Especially since you're the new reigning champion! Thank you!

Interesting, BIOS update mentions SSD performance:
http://www-307.ibm.com/pc/support/site. ... MIGR-70802

Erik and tinkererguy, I ran the benchmark from the link Erik posted but I have no idea how to interpret it or how to post an image of the graph.

Here's three datapoints from the benchmark using my RAID 1 Seagate drives.

Block Size Write Read

smallest
0.5 2801 2199

medium
512 68217 63535

largest
8192 71392 74981

Barry

barrywohl wrote:Erik and tinkererguy, I ran the benchmark from the link Erik posted but I have no idea how to interpret it or how to post an image of the graph.

Here's three datapoints from the benchmark using my RAID 1 Seagate drives.

Block Size Write Read

smallest
0.5 2801 2199

medium
512 68217 63535

largest
8192 71392 74981

Barry

Thank you for posting, Barry! I didn't mean to make this harder than it needs to be. Hopefully this will help. If you can eyeball the approximate average Write and Read thruputs for 64KB - 8192KB transfers, then post here. So, take the red Write lines, and look at the x axis in MB/Sec, and estimate the average, since they're all likely similar anyway. Then again for the green Read lines, here's the screenshot again for reference:
http://picasaweb.google.com/tinkererguy ... 0405449714
Then perhaps edit something like this format below as a template, that'd be great, saving the trouble of screenshots. Here's a synthetic example, based on guess-timates from your posted data:
==========
OS: Vista x64
Model: W700 Quad Core 2.53GHz
Settings: Device Manager, Drives, Drive Properties, Policies tab
Optimize for performance (radio box checked)
"Enable write caching on the disk" checkbox on
"Enable advanced performance" checkbox off
Antivirus disabled
Tool: ATTO Disk Benchmark v2.3 32 bit
Approximate average for 64KB to 8192KB transfers:

C: 60MB/Sec Write, 60MB/Sec Read
W700 with RAID1 [BRAND MODEL SIZE RPM of the drives]

D: 80MB/Sec Write, 80MB/Sec Read
W700 with UltraBay SSD [BRAND MODEL SIZE RPM of the drive]
==========
I realize this scheme may be flawed, and perhaps somebody has a truly free (not trialware) tool to suggest that is simpler to post results from, I'm all "ears." Best would be text output that we could cut-and-paste here to minimize the effort required.

OS: Vista x64
Model: W700 Quad Core 2.53GHz
Settings: Device Manager, Drives, Drive Properties, Policies tab
Optimize for performance (radio box checked)
"Enable write caching on the disk" checkbox on
"Enable advanced performance" checkbox off
Antivirus McAfee preinstalled. not disabled. (I forgot)
Tool: ATTO Disk Benchmark v2.43 32 bit
Approximate average for 64KB to 8192KB transfers:
Write speed 66852 (average of 8 block sizes 64 to 8192)
Read spped 62283 (average of 8 block sizes 64 to 8192)
Seagate Momentus 7200.3 ST9320421AS two mirrored RAID 1 drives
C: 67MB/Sec Write, 62MB/Sec Read
W700 with RAID1

I would doubt the accuracy of ATTO benchmark I remember an article where ATTO has yielded very unusual benchmark results... If you both really want to test the read performance of your RAID setups, then I would recommend a more "conventional" tool like HDTach or HD Tune. Unfortunately, the free versions cannot test write performance (note however that these tools are for now not suitable for testing SSD performance).

Just my 2 cents

Marin

tinkererguy wrote: Here's the previous test, C: internal 200GB and D: Ultrabay 320GB:
==========
OS: Vista x64
Model: T61p
Settings: Device Manager, Drives, Drive Properties, Policies tab
Optimize for performance (radio box checked)
"Enable write caching on the disk" checkbox on
"Enable advanced performance" checkbox off
Antivirus disabled (Symantec 10.2 x64 on or off, got same results FYI)

Tool: ATTO Disk Benchmark v2.3 32 bit
Approximate average for 64KB to 8192KB transfers:

C: 51MB/Sec Write, 55MB/Sec Read
T61p Vista x64 C drive Hitachi 7K200 HTS722020K9SA00 200gb 7200RPM

D: 15MB/Sec Write, 14MB/Sec Read
T61p Vista x64 D drive Western Digital Scorpio WD3200BEVT 320GB 5400rpm in Ultrabay

S: 55MB/Sec Write, 54MB/Sec Read
T61p Vista x64 S drive Western Digital SE16 WD7500AAKS 750GB 7200rpm on eSATA SI3132 PCI Express card in docking station
Thermaltake enclosure

F: 93MB/Sec Write, 97MB/Sec Read
T61p Vista x64 F drive Western Digital WD10 EACS-65D6BO 1024GB (7200rpm?) on eSATA SI3132 PCI Express card in docking station
Calvary enclosure
==========

Tried this test again today, this time with T61p internal bay with Intel 80GB SSD X-25M C: drive in Ultrabay, and Western Digital 320GB D: drive 7200 internal, with the same data/boot OS as the previously tested configuratino (cloned with Ghost 11).

Interesting results, still don't know why Ultrbay showed so slow before, much better overall system performance now too, see also screenshots:
http://picasaweb.google.com/tinkererguy ... 6733657794

==========
OS: Vista x64
Model: T61p
Settings: Device Manager, Drives, Drive Properties, Policies tab
Optimize for performance (radio box checked)
"Enable write caching on the disk" checkbox on
"Enable advanced performance" checkbox off
Antivirus disabled (Symantec 10.2 x64 on or off, got same results FYI)

Tool: ATTO Disk Benchmark v2.3 32 bit
Approximate average for 64KB to 8192KB transfers:

C: 55MB/Sec Write, 83MB/Sec Read
T61p Vista x64 C drive Intel SSD X-25M 80GB in Ultrabay

D: 50MB/Sec Write, 50MB/Sec Read
T61p Vista x64 D drive Western Digital Scorpio WD3200BEVT 320GB 5400rpm internal
==========

tinkererguy wrote:QFoam, if you could let us know where you bought your rail kit, it'd be appreciated. Thanks!

tinkererguy, sorry to take so long to get back to you, I've been out with a cold.

Anyway, it sounds like you did just fine with the rails/caddy. I've put up a web page with some info about rails/caddys for the W700:

http://www.buy-a-thon.com/reviews/W700- ... -caddy.htm

and will be adding to it the distributor you mentioned, plus photos of installing my new hard drive/rails/caddy into my new W700 (which I'm just about to do). That's linked to a page I just created which takes about 160 pages of notes I took while researching my W700 purchase, and boils them down to one page of links:

http://www.buy-a-thon.com/reviews/W700-resources.htm

If you think of anything I should add, let me know (there's a suggestion box for the page). I figure that, if I could have found a list of links like that, it would have saved me a heck of a lot of time.

Such good stuff you wrote up, amazing really.

I guess good things do come to those who wait, a wait well worthwhile!

Just ran another test, this time with SSD internal and 320GB in Ultrabay, again no slowdown noted, other than the "usual" 50MB/sec cap that Ultrabay seems to enforce. This time, the SSD really is able to "fly", in the internal drive bay, check out those speeds, can't wait to try on W700!

Here's the new pix,a dded to existing W700 pix:
http://picasaweb.google.com/tinkererguy ... 9325276834

==========
OS: Vista x64
Model: T61p
Settings: Device Manager, Drives, Drive Properties, Policies tab
Optimize for performance (radio box checked)
"Enable write caching on the disk" checkbox on
"Enable advanced performance" checkbox off
Antivirus disabled (Symantec 10.2 x64 on or off, got same results FYI)

Tool: ATTO Disk Benchmark v2.3 32 bit
Approximate average for 64KB to 8192KB transfers:

C: 53MB/Sec Write, 105MB/Sec Read
T61p Vista x64 C drive Intel SSD X-25M 80GB internal

D: 50MB/Sec Write, 51MB/Sec Read
T61p Vista x64 D drive Western Digital Scorpio WD3200BEVT 320GB 5400rpm in Ultrabay

==========

tinkererguy -

not that we don't appreciate your tests but you're getting off-topic here.   the point of your thread is to point out speed results from your W700, not a T61p.

if you want to post tests from a T61p, i suggest you start a new thread to keep this one from going further off-topic.   users looking for W700 info will be confused when all of your tests above are from a different machine.

thanks for understanding.

QFoam wrote:Anyway, it sounds like you did just fine with the rails/caddy. I've put up a web page with some info about rails/caddys for the W700:

http://www.buy-a-thon.com/reviews/W700- ... -caddy.htm

fyi, both the parts call center and parts warehouse are in pennsylvania, not colorado.   the call center is in mechanicsburg and warehouse is in carlisle.

erik wrote:fyi, both the parts call center and parts warehouse are in pennsylvania, not colorado. the call center is in mechanicsburg and warehouse is in carlisle.

Thanks for that info, Erik. I called IBM's Parts Center today. They told me that, despite their search page saying they're located in Boulder, CO, it's out-of-date and they moved to Mechanicsburg, PA, about 8 years ago.

The parts I recieved were indeed shipped from Carlisle, PA, although the person at the Parts Center said they ship from other locations, for example Georgia or Illinois, if the part is out of stock in PA.

So I've corrected the information on my rails/caddy page to reflect all of this, and appreciate you pointing out the error.

QFoam, your W700 site is amazing (finally had a chance to read thru the whole thing), and I'm flattered to have some of my posts linked to.

One very minor thing, a footnote really. You say: "But previous-generation laptops were instead limited to a supported maximum of 4GB of RAM", but actually it turns out you can get 2 4GB sticks in a T61p and it sees all 8GB or ram in Vista 64 just fine (running that config right now myself). But true enough, it may not be supported. More here:
http://blogs.technet.com/keithcombs/arc ... f-ram.aspx
http://forum.thinkpads.com/viewtopic.ph ... 10772fcf0a

Which reminds me, Keith Combs blog will hopefully cover the W700 someday soon, see here:
http://blogs.technet.com/keithcombs/arc ... order.aspx

tinkererguy wrote:One very minor thing, a footnote really. You say: "But previous-generation laptops were instead limited to a supported maximum of 4GB of RAM", but actually it turns out you can get 2 4GB sticks in a T61p and it sees all 8GB or ram in Vista 64 just fine (running that config right now myself). But true enough, it may not be supported. More here:

I think Erik's going to throw us into a new thread any moment now, but you're right. And your guess about why I said "supported" is correct. I'll put a note to that effect in that section.

But, getting back to this thread , the support for 8GB is an important performance advantage of the W700 over the new Mac laptops, for example. Apple only supports 4GB in the new machines, although users have successfully used 6GB (with the limiting factor probably being Apple's EFI, their equivalent of our ancient BIOS). So with the W700, that means you can rely more on that 8GB of RAM, and less on virtual memory and swapping to disk.

And that impacts how you'll set up your high-speed storage system. Your 7200 RPM Seagate RAID-0 is probably going to have much faster transfer rates for writes than will your SSD system drive. So should you devote the fastest part of your RAID-0 to swap space, or will the extra memory let you get away with leaving the swap file on the system drive?

That's one thing that would be interesting to see in your tests. Granted, you'll have 6GB of RAM instead of 8. But how much of a performance hit will you take, for various computing tasks, by leaving the swap file on the SSD? For example, a lot of people are going to be using their W700s for manipulating large image files in Photoshop CS4. Will the drive on which the swap file resides significantly affect performance in that case? (I also wonder whether accessing a large data file from a drive, while swapping to that same drive, will reduce performance due to contention for the drive.) On the other hand, as I recall, you're going to be doing a lot of virtualization. Does having the swap file on the SSD instead have a much greater peformance impact on that type of task?

But thanks for your input regarding the memory limits. I'll put in links to Keith Combs' efforts. I still have a large stack of changes to make and things to add.

Ah yes, keeping me on track, thanks QFoam and Erik.

My rationale for posting T61p SSD results were for comparisons to W700, giving shoppers the kind of metrics they might find handy. I was asked a reasonable question, that is, how much of the boost is due to SSD and or RAID0, and how much is due to merely moving from T61p to W700? Well, hopefully soon, I'll be able to share that info, once I get my hands on my W700 again. And now that my T61p has the C: drive SSD internal, moving to W700 soon, it'll be more apples to apples.

Meanwhile, I was also curious to try to see what folks get with disk speeds on their W700s, comparing RAID1 with RAID0, and/or different brands of SSDs too.

And QFoam, you asked about VMware. I generally leaveVMware Workstation set to not start a VM unless real RAM can be made available (rather than paging), so in that case, placement of swap file probably not important.

FYI, as I try to watch drive IO during long tasks, like rendering Camtasia videos, I will probably go with watching drives using the HDDLed activity monitor, to help me keep tabs on disk I/O and queue depth as well. See http://www.hddled.com It can be a little flakey at times on x64 Vista, but I've found nothing like it, for monitoring things like how much I/O was going on to/from ReadyBoost in the SD slot, versus USB 2.0 key, etc. Now that I'm at 8GB on T61p I don't bother with Readyboost anymore.

I think "cookbook" at the start of this thread is as cooked as it needs to be at this point, I basically edited it heavily, after incorporating feedback from you all. Thanks!

New pictures/results posted for W700 drives posted, see screenshots here, C SSD and D RAID0 respectively:
http://picasaweb.google.com/tinkererguy ... 3334580434
http://picasaweb.google.com/tinkererguy ... 7256352066

and summary below. Note, I also tried with Intel ICH9M RAID0 for the 2 320GB drives, and got very similar results (with 10%), but I don't have good screenshots of that, and the W700 would hang at boot if docked with an eSATA attached (BIOS couldn't see UltraBay drive as boot target for some reason, turn off eSATA drive, it came back and booted). So I'm using Vista created RAID0 D: using Dymanic Volumes with Disk Manager, and the CPU burden seems minimal and the speed is excellent and backup/restore works fine (using Windows Home Server).

Also, the Intel SSD X-25M C: boot drive "feels" much faster than benchmarks would indicate, with dramatically faster day-to-day operations. See a spreadsheet I created here for a better idea of the practical speedup here:
http://picasaweb.google.com/tinkererguy ... 2489042178

==========
OS: Vista x64
Model: W700
Settings: Device Manager, Drives, Drive Properties, Policies tab
Optimize for performance (radio box checked)
"Enable write caching on the disk" checkbox on
"Enable advanced performance" checkbox off
Antivirus disabled (Symantec 10.2 x64 on or off, got same results FYI)

Tool: ATTO Disk Benchmark v2.3 32 bit
Approximate average for 64KB to 8192KB transfers:

C: 57MB/Sec Write, 160MB/Sec Read
W700 Vista x64 C drive Intel X-25M 80GB in UltraBay

D: 123MB/Sec Write, 140MB/Sec Read
W700 Vista x64 D drive WindowsDynamicVolumeRAID0 WD 320GB ST9320421AS 7200rpm internal

==========

One remark to using 7200 rpm drives for the RAID 0: If you compare the performance charts of a 320GB WD Scorpio Black (7200 RPM) to the Scorpio Blue model (500GB, 5400 RPM), they turn out to be almost identical. This is because a 500GB HDD has a much higher data density, which compensates the higher rotation speed of a 320GB drive. Thus if HDD space is an issue I'd recommend to take the bigger drives with no significant loss in performance.

In addition I have one question about switching from a single HDD to RAID 0. If I install a second identical drive and change BIOS settings from AHCI to RAID 0, will the data on the former single drive be erased, or is it automatically expanded to the 2nd HDD without any data loss?

Thanks

muol wrote:One remark to using 7200 rpm drives for the RAID 0: If you compare the performance charts of a 320GB WD Scorpio Black (7200 RPM) to the Scorpio Blue model (500GB, 5400 RPM), they turn out to be almost identical. This is because a 500GB HDD has a much higher data density, which compensates the higher rotation speed of a 320GB drive. Thus if HDD space is an issue I'd recommend to take the bigger drives with no significant loss in performance.

In addition I have one question about switching from a single HDD to RAID 0. If I install a second identical drive and change BIOS settings from AHCI to RAID 0, will the data on the former single drive be erased, or is it automatically expanded to the 2nd HDD without any data loss?

Thanks

On your first point, I agree, I just happened to already have one 7200 320GB drive, so matching it with a second one for RAID0 made sense, as well as the favorable reviews. When another comprehensive review comes out for 2.5" drives greater than 500GB, similar to the review I referred to on 7200 rpm 2.5" drives:
http://www.tomshardware.com/reviews/not ... ,2006.html
and the price drops a bit, I might make the jump to that capacity, likely 2Q09 or later, as > 1 Terabtye would be preferred to the 595GB formatted capacity I have now, of course. And I can always use the 320GB drives to upgrade my families T61p's.

On your second point, I'm nearly certain that when I was playing with RAID, it not only warned about data deletion, but actually did kill access to all the data as well, immediately. Then as new data was laid down, in my case, restoring a backup of my D: volume from Windows Home Server, that data is striped across both spindles roughly 50/50, overwriting whatever was left on both of those spindles.

But I also found it annoying that I had to turn off a dock attached eSATA drive to get the BIOS to recognize the boot drive (strange behavior, present in event the latest BIOS). So I gave up and went back to AHCI mode, and did the RAID0 striping from within Vista x64, with apparently no degradation in peformance:
http://picasaweb.google.com/tinkererguy ... 7256352066