Sony Vaio Z i5/i7 Official Owners Thread

If I had to bet on either of:

Sony having pulled a fast one, knowing that there's one born every minute

or

Sony having made a breakthrough invention that solves the SSD+RAID=No-TRIM problem without marketing it and milking it for all it's worth

... well, I know where I would put my money.

 

Haha, yeah the PS3 is ample proof of that (dual 1080p output and certain PS2 backward compatibility anyone ). But they do target a premium market with this new Z, could they really get away with stuttering SSD's without paying dearly?

 

Hmmm. Don't recall either the 10+ hours thing or the CLIE thing from those people.

 

And, of course, if the former, betting against the odds of a class-action suit.

And, while the latter does indeed stretch credulity, the Z itself represents something no one else in the world has been able to accomplish so far in terms of packaging, power, etc.

 

I don't think all those steps are necessary for writing to a saturated SSD, provided there are free pages in the spare area. All SSDs use a table to map logical addresses to physical pages. It may be in RAM like you said, but it also has to be persistent. So maybe it's battery-backed and flushed to the SSD itself upon loss of power or maybe expensive nonvolatile memory is used to store the table, I don't know. But since this table is needed for the SSD to function at all I think we can assume that SSD makers have solved the problem of how to keep this table persistent and protected against power loss.

With the block table, no free blocks in the main area, and free blocks in the spare area, writes can go as follows:
- Calling process issues write command to LBA X
- Write data to page A in spare area
- Mark page in main area currently pointed to by LBA X as unused
- Remap LBA X to point to page A.
- Tell the calling process the write is done
- Block in spare area that contains page A now becomes part of the main area, remember to reclaim a block for the spare area later.

The spare area allows you to write to a saturated drive as if it has free blocks. There's no need to copy an entire block to the spare area with the overwritten pages and copy back to the main area, you can just remap the block in the block table and let the block in the spare area become a part of the main area. Of course this is not enough, that's where GC comes in...

Before and after this write, the amount of over-provisioned space should be invariant. The number of pages written to in the spare area should be the number of pages that are marked as unused in the main area. Background GC can then consolidate the blocks that were marked as having free pages, thereby freeing up the same number of blocks that were used up in the spare area. These blocks are added back to the spare area, so that the number of free blocks in the spare area is invariant after garbage collection.

This should mean that if the size of the data to be written in a given time is smaller than the size of the spare area, and if after the writes enough time is given to allow the GC to reclaim as many spare area blocks as had been lost, there should be little performance loss. But of course if the writes exceed the size of the spare area or if writes come in while garbage collection is in progress we are back to saturated mode writing.

This sounds to me like with a generously overprovisioned drive and good garbage collection, write performance losses can be avoided in most cases even after saturation. I'm unsure about some of these details, so I guess as usual let me know if this doesn't work.

 

Here's the thing...guys like me wouldn't know about TRIM without guys like you. I would've bought the Z and never would've known why it slowed down. How many guys like me are there buying laptops and how many guys like you are buying laptops. I think the group with guys like me is much bigger.

 

Just ordered the new Z (CTO) from Sony in Germany.

First thing I'm gonna do is re-use my microSATA adapter I made for the old Z, for replacing Optical drive with X25-E

Question - I assume new Z requires new docking station. Is that available somewhere?

 

Question - I assume new Z requires new docking station. Is that available somewhere?[/QUOTE]


Model number: VGP-PRZ10 Series Docking Station

 

+1

Intel's "Matrix RAID" technology is basically software RAID, with hardware BIOS support. "Faker-aid" is a good name for it, but I see it referred to as SoftRAID. This is basically the same thing Nvidia does on their motherboards as well (forgot what marketing name they have for it).

SoftRAID is generally fine for simple RAID types like 1 (mirroring) and 0 (striping). RAID 0+1 is probably OK too with this setup.

When you do more complicated configurations like RAID5, you will likely take some type of a performance hit. These RAID types need parity calculations to maintain the array's integrity. True hardware RAID controllers offload these calculations from the system CPU and can offer dedicated cache to disk based operations. We'll have to wait and see what RAID5 does to performance once somebody configures it. Might be minimal

Both Faker-aid and true hardware controllers can hide the fact that multiple hard drives are being used to the OS. This simplifies usage for users who don't want to modify things or just do normal work.

 

Can be found for ~170 EUR in Germany.... Oh well.. I guess I'll recover some money from selling the old docking station on Ebay.

I dunno for others, but I always have bittersweet taste when buying Sony products... new power and features are sweet, but you always have that aftertaste of being ripped off for everything else (new docking stations, new batteries, ...)

 

Why? I haven't seen Sony make any claims for random write speeds.

This is much like a phone company or cable provider selling you an internet connection which is "up to 6 Mbps". They haven't promised you that you will always get that, and they haven't said anything about the upload speeds. That it's really 0-6 Mbps down and 0-768 kbps up hasn't triggered any class action lawsuits.
Those who can't live without a guaranteed minimum rate speed (or even know what that means) buy their lines elsewhere. But they're in minority, so the phone companies and cable providers still sell well.

I am fairly certain that Sony, like all big coprorations, have lawyers that go through all their marketing materials, and remove claims that might get them into trouble, while preserving as much hype as possible.

 

Which is precisely why Sony would be taking a risk, possibly calculated, that they would be creating a potential firestorm for themselves if they weren't cognizant of this issue ahead of time.

 

The drives don't map individual blocks, but entire sectors.
So when you write your single block, the drive still has to concatenate it with the surrounding data, and can't erase just that one block.

Normally, they use two NVRAM areas for the mapping (so they can write to one and keep the other one ready for writes, without the dreaded erase delay).

A solution to the problem that allows something similar to what you describe without losing data integrity no matter when a power outage occurs is to set aside a journal area in the "spare" area. The journal contains a record of which operations are still pending (much like the NTFS journal does for the file system). That way, if power is lost, the journal can be replayed, and a consistent state reached. However, as it adds another synchronous write step for each write operation (which slows them down compared to the competition), it's not functionality you find in drives marketed for their raw speed.

 

As a matter of fact, they clearly promote "6 times" faster than HDD.

 

arth1, you are extremely knowledgeable and I pretty much agree with every info and advice you provided on this thread except for this part about RAM disk usage.

When I use a RAM disk with an SSD, it is not used for caching per se, although it is a form of caching, but it's to direct temporary data that do not need to be written to the SSD. RAM disk for temp file/folder usage has been tremendously useful particularly for lessening unnecessary writes to the SSD. Not only do I use it for Windows temp folder, but also for Firefox's temp folder, and other programs such as Nero and a few others that does not come immediately to mind.

The last statement from the quoted paragraph is true - I do use it for loading some programs that I want instant access to. But it's not primarily for that.

 

Your brain must have skipped the "up to" or footnote explaining just what they mean.
No way would Sony (or anyone else) make a blanco claim about the speed.

 

I think you just proved arth1's point. From the Sony press release:

(bolding mine)

No way would Sony claim minimum 6x speed gains with quad raid...

Edit: arth1 beat me to it. Funny how our last sentence is almost identical...

-Peter

 

Haha, true! I haven't heard of any software though, maybe they will in the future. Does Sony have a history of supporting old hardware?

 

You're correct -- for temporary files, it can serve a purpose in keeping them off the disk altogether.

I'm spoilt by using xfs as a file system, which has the unique feature that short-lived temporary files never will be committed to the disk at all. So I seldom think about this as a problem -- for me, it would only apply to long-lived temporary files, and those, I do want flushed to disk, so the RAM can be used for caching other objects instead.
But yeah, if I set up a system with ext2/3/4, I generally make /tmp a dynamic RAM disk. With xfs, I don't.

 

No.

Though I agree with rmcx that people on this forum are being a little bit overzealous regarding the situation over the SSDs. Sure, we know that having the drives RAIDed most likely makes TRIM unavailable. However, we still haven't confirmed the situation regarding JBOD and TRIM. Which would be the ideal solution. The latter situation doesn't look too improbable to me, so I don't understand why some can be so conclusive.

 

According to someone in the preorder thread, the Sony Style in South Coast Plaza has the new Z's. I'm only 10 minutes away from it, but I have a final tomorrow, so I can't go until Saturday.

 

ummmm...magicgate...it is STILL on the Z

 

If I feel I've studied enough for my test, I'll go today rather. However, I'll need someone to tell me what I need to do before I go, such as checking the SSD stuff and whatnot.

 

True, but the keyword here is "promote", and that is often enough to get lawyers interested.

I'd have to look, but I believe that there have been suits about promoted broadband speeds in the past.

http://interact.stltoday.com/blogzo...aces-class-action-suit-over-broadband-speeds/

http://www.dslreports.com/shownews/HughesNet-Sued-For-Poor-Service-102542

http://en.wikipedia.org/wiki/Clearwire

 

That's good to hear because I am driving down from the central coast and I called the Camarillo Sony Style outlet, and the guy claimed they will not see it for 6 months at least.

 

I can understand what you're saying.

My folks definitely fall into that category. If they are to buy the VAIO Z & the SSD slows down in the future, the 1st thing my dad will tell my mom something like "Oh honey, our notebook has slowed down & it cannot really handle our application programs anymore, I think it's time to buy a new notebook." Since my folks will most porbably have quite decent experiences with the notebook (except the eventual slowdown, if any), their replacement notebook purchase will still be a Sony VAIO & guess what? Sony will most probably be "laughing all the way to the bank" over ignorant customers like my parents!

On the other hand, I think further down the road, there will be solutions to eliminate most of this SSD slowdown problems through a SSD firmware update or something.

 

This is from Sony UK's Z page:

With the innovative Quad SSD Raid drive (*), you’ll be able to work faster and quieter than ever before while transferring data at over 6 times the speed of a normal hard disc drive. Saves time and enhances your productivity.

(The asterisk refers to "selected models").

I'm not sure I'm seeing any equivocation there.

 

IMHO, no matter what one buys (whether it is a Sony product or not), it is always good to take the advertised specifications & "glorified" marketing information with a grain of salt. The best way to know & feel how good the suggested item is is to actually buy it home & use it for a period of time yourself. Of course, this depends on the kind of return policy the manufacturer/retailer gives you & whether you can afford to pay for the item. Good luck.

 

But really, now, who would spend 4-6 times the price of an "average" laptop that would fall into this category? It's not like you can walk into Best Buy and pick one of these babies up. You can barely get them at Fry's. I suppose you could walk into Sony Style and do it...some day!

The Y model, for instance, sitting right next to that Z, is about the same size and 1/2 the price.

I think the argument that this is Sony trying to ding the "average Joe" is just a bit specious.

 

I definitely agree with you on certain portions of your comments.

The thing IMHO is that if one is to spend like $3,000-$4,000 on a notebook, wouldn't it be in the best interest for that person to at least put in some serious time & effort in researching for what he/she is buying? No offense, I mean you don't just throw such a large amount of money into something just by fully believing everything the marketing literature hypes up, do you?

By the way, the SonyStyle online store offers a fabulous 30-day return policy, which one can make good use of, if needed, IMHO.

 

Why is it so hard to believe that Sony would screw us over? As said, its not illegal considering they didnt market anything that was untrue.


Normal consumers arnt going to know the dangers of Raid and SSDs with no trim.

 

Fully agree with you on the try/return issue. I rarely do any major (>$20???) purchase that I can't reasonably return. And, yes, in this way I am willing to jump in at the $3-$4k range.

(I bought an HP Envy 15 a few months ago. Took it home, tried it for a few weeks, returned it. HP offered me a discount to keep it. I asked "how much" and they said "$100"...on a $3000 unit! I said "keep your money and your computer.")

 

IMO, Sony is still regarded as the leading brand. I suspect many people trust the Sony brand And besides who is going to check out whether SSD has TRIM in the real world. Who is even going to think of that? Gosh, it even took you experts dozens of pages of discussion before the subject even came up because everyone was "wowed" by the power and form factor. A buyer goes to the Sony Style Store, (I know a young Doctor who didn't even know you could CTO), the salesman tells him SSD is state of the art, with Sony quality backing it, the buyer says give me at least 250gb, and off he/she goes. Listen people are strugglng to keep up with all of the information, especially in their professions, who has time to learn it all. There are only so many minutes in a day.

 

I've read this TRIM discussion with some amusement and considerable frustration for a long while now and have a simple question:

Will the full extent of any TRIM support (or other workable solution) be known in a day, two days, a week, or at least w/in 30 days of Arth1 and others "in the know" receiving their Zs? I'd love a clear sense of whether there is any meaningful solution w/ minimal fuss (e.g., no reconfiguration of the RAIDs) w/in the 30-day return period.

Thanks.

 

I just want to make it clear that I do not in the least discount the value of the information that's available in both this thread and the previous one. Nor do I discount the efforts of most here who are trying to learn and understand what's what.

My issue is that now that we know what we should be looking for, shouldn't we be basing our decisions on what the thing actually does as opposed to what it might and/or might not do? And for this, while no one really has any data points, some are willing to make a commitment that may well be hard to undo in the future.

My suggestion is that people who are reticent about being first into the pool should hold off another month or two until the reviews are in rather than potentially cutting themselves off at the knees right at the beginning.

 

Reverend!

More of this, pls

 

The final solution is for Intel to release an Intel RST/Matrix driver with TRIM support (assuming the Samsungs are TRIM enabled which they surely must be).

In numerous discusions with Intel tech support they have indicated to me that they are working on it, but who knows how long it will take ....

 

But it doesn't work. I have an old 64 MB card with ATRAC3 music files on it, but I can't access it from my Z. SonicStage won't authorize the drive.

So I would have to say no, Sony does not support old hardware.

 

No. For a multi-thousand-dollar purchase you bleeping well should base your decision on how it will behave in the future.

FACT: The Z as delivered does not support TRIM
FACT: Drives without TRIM saturate and experience stalls.
FACT: Sony does not have a track record of providing updates.

On the other hand, a hope that Sony for some reason has developed a solution for this is pure speculation. Yes, I would like to base my purchases on what we know, not speculation and wishes.

 

The phrase "what the thing actually does" is meant to include its entire useful life span. Sorry if that wasn't clear. And your "facts" still do not include any real-world knowledge about this particular unit's current and/or potential performance.

I welcome anyone who can provide me with those kinds of numbers as truly, I do wish to maximize a multi-thousand dollar purchase. As a friend of mine once said: "Numbers, give me numbers".

 

Take another look at what they actually say:

WHEN you transfer files at over six times the speed of a normal HD,
THEN you will work faster and quieter than ever before.

In other words, what comes after "while" is a premise, not a promise.

 

I do not have any real world knowledge that says that if I drop my Z from the Eiffel tower, it will break either. That doesn't mean I shouldn't think it to be true.

Based on empiric evidence of how SSD drives behave, it should be expected that the SSDs in the VPC-Z will behave the same, until and unless the opposite has been shown.

 

I believe your interpretation here is a bit of a stretch and in any event, not something an end user is likely to parse at that level.

 

According to this paper: http://portal.acm.org/citation.cfm?id=1534544 the drives actually do map individual pages, or even fractions of pages. The generic NAND SSD architecture in the paper uses an out-of-place write and page-level garbage collection mechanism exactly as I described.

the LBA block size used in the paper is actually 512 bytes, so the granularity of the mapping is down to a fraction of a flash page. With the out-of-place write mechanism as described, as long as there are free pages in the spare area of the drive, even after saturation, writes proceed as normal and there is no need to concatenate new data with existing data in the overwritten block.

 

I think this SSD issue has gotten a little out of control..... Notebook manufacturers have ALWAYS put the cheapest slowest performing SSDs in their laptops, it just makes good business sense. But enthusiasts keep trying to expect intel x25 performance from OEM SSDs, it just is not going to happen. PERIOD.

Samsung's have been used in Macbooks for years now, and are consistently "bad" peforming SSDs, however they still just manage to be better than traditional HDDs, even when saturated and without TRIM. See the following article.

http://www.anandtech.com/storage/showdoc.aspx?i=3631&p=19

Now it looks like Sony has used basically the same style SSDs (ie cheap), however in RAID 0 to push a little more performance out of them. These RAIDed drives therefore should still outperform a traditional platter HDD in all situations, even when every LBA has been written to, and without TRIM, their 4k random reads/writes should still not produce stuttering as was seen in the earlier Jmicron controlled SSDs.

The lack of TRIM (I think this is pretty much a given, since NO OTHER SSDs in the world today support TRIM while RAIDed) will obviously mean that the drives performance will degrade over time, however even at maximum degradation (see the above article) the drives should still be slightly ahead of a very fast traditional HDD.

Inexperienced users should not be put of by all the "enthusiasts" who want the best of the best of the best at all times. Yes you will not have TRIM, but likely you should not notice any difference in typical usage, compared to any other HDD based notebook.

**DISCLAIMER** of course all discussions will end up being basic conjecture, and based on research of similar products, as the actual number of units out in the world is quite limited, feel free to comment and tell me I am wrong!

 

Ok, found this thread of someone with an SSD RAID0 setup (not on a Z, I don't think) that experienced degradation, and the different methods he used to finally get performance back to normal. Interesting that the tony-trim method seemed to make matters worse.

Another thread confirming this method works. It goes into a bit more detail (including the need for the Win7 DVD to reinstall the Windows backup image).

-Peter

 

Very encouraging.

 

Those of you who have the ssd (128, 192 etc.), what is your solution for storage? At home, do you have it connected to a external harddrive and use that as the storage? What is the loss of speed by choosing this method? 128 gig which I will get is not nearly enough for my media, work. I will only have the os, apps on the ssd and will need a bigger solution for my storage purposes.

 

I've been able to confirm that, at a minimum, the 64GB Samsung SSDs support Trim.

Jury is still out on RAID and Trim. Diskinfo didn't work for me, it couldn't find a drive. It found the drive when I broke the RAID volume and installed Windows cleanly



I'm running the recovery CD's to revert back to factory configs. I'll give CrystalInfo another try shortly.

 

I've posted a ton of pics and am about to add some videos in this thread:
http://forum.notebookreview.com/showthread.php?t=466881