SSD and Windows XP

I'm using Windows XP (won't upgrade until later this year), but I need a good SSD drive to improve performance, I read somewhere that since XP doesn't support Trim, I need to consider other than just the newest drives, is this correct?

I've looked at Intel 510 series and Intel 320 series, what do you recommend? Or should I look at other alternatives?

 

Get a drive that has/supports garbage collection (sometimes called GC).

 

Isn't that the same as TRIM ?

 

No. TRIM is just a(n extra) command that lets a SSD know when a file has been deleted. Garbage collection is what does the actual deleting and cleaning up of used flash pages. So, while they often work together, they're not actually the same thing.

Garbage collection, for example (from my understanding) can still work perfectly fine without TRIM. Even though the drive doesn't know that certain data blocks have been deleted, and thus can now be cleared, what it can do is take all the writes that are typically spread over multiple pages (because that's how SSDs achieve their high write speeds, writing to multiple pages at once), and then consolidating them into a few "full" pages, thus turning several partially full blocks into fewer completely full blocks, leaving more "empty" blocks to be written to (and thus keeping speeds close to "clean" state). TRIM can help with this, because it lets the garbage collection know which data can actually be completely deleted (and not just consolidated), which improves the efficiency of garbage collection.

 

If you're upgrading just the O/S in a few months: get the Intel X25-M 160GB or 120GB models.

If you're upgrading to a new system (with SATA3 and SNB) I would be only considering the 250GB 510 right now. (It will give relatively worse performance on your current setup, but you will be ready to move it to the platform it was meant for - it should still be equal to or faster than the X25-M though...)

Keep in mind that there are hints that a 700 series (Intel) is coming too. This may be a reason to simply buy the smaller/cheaper 120GB X25-M model now - but you will be gambling with the timing of the availability of the 700's and your new (possible?) purchase.

Good luck.

 

So, and old SSD from Intel will perform better than the new SSD when using Windows XP, is this correct?

We will upgrade to Windows 7 eventually, but we need the best performance boost now with Windows XP.

Thanks

 

Do you consider the Intel X25-M old?

 

Yeah well, I see them disappearing from the online stores in my country, the new are the 510 and 320 series.

 

Probably not. Performance will probably still be better with the new SSDs.

 

Wasn't one of the Kingston models developed for Apple with very strong GC, so that Apple didn't have to implement TRIM?
I think it was the V100 or the V+100.
(but that would be overkill on Win7)

 

Apparently. Kingston SSDNow V+100 Review - AnandTech :: Your Source for Hardware Analysis and News

 

Kingston's SSDs have performance levels comparable with what the competition was offering 2 generations/years ago. If this were April 2009 and the only other options were Intel and Indilinx, they'd be worth considering.

 

you switch in a few month? just get a proper ssd with trim and all for AFTER your switch. xp will work just fine till you switch.

 

Why nobody mentioned that IntelToolbox sends EXACTLY Trim commands to SSD (sheduled) on systems without TRIM (like XP, Vista-that I'm using) ?! Problem solved.

 

Chris ast1,

Because that only looks at one possible solution: simply keeping the performance as high as possible in XP.

If the OP buys a new system/platform with SATA3 support in a few short months - buying a SATA2 SSD is not in his best interests.

If he/she buys a newer system (w/Win7), but still limited to SATA2, buying a SATA3 (even Intel) SSD is again, not in his best interests.

So while Intel SSD Toolbox solves a single issue, it cannot account for the others being discussed here that are as important (if not more so).

 

Ok. Point taken. But isn't it true that SSDs with faster Random 4KB R/W are faster than drives that just declare fast sequential speeds? And, did You notice any drive on the market that SATURATES sata 2 with his Random 4KB R/W or IOPS speeds ? Because as I'm aware there's no such drive ... Vertex 3 is closest with it's Random write speed 60MB/s. Most SSD's deliver 20 - 50MB/s in this scenario. With queue depth set to 32, they deliver 120/130MB/s.

 

No, it's not true that faster 4K R r/w's dictate (solely) a drives speed.

Many many more considerations before the 4K R r/w 'scores' become relevant for normal users. 4K R r/w's are targeted for servers - workstation users don't use computers like that.

 

4K writes don't matter too much for normal users, sure, but 4K random reads sure do! Open a file? Yep. Small random read. Launch a program? That's a flood of small random reads (plus some big sequential ones). Multitask at all? Small random reads. Etc.

 

I think we need to point out here that "fast" and "slow" are all relative - most of what you'll be doing are small, random reads, which even a flash drive is better at than a HDD (hence ReadyBoost's existence). I went from a HDD to Sata 2 SSD in my tablet, and the difference is night and day. Boot time went down by at least 50%, the desktop is actually usable as soon as it appears, and programs load pretty much instantly.

 

True. Just about any non-JMicron SSD will be orders of magnitude faster than a mechanical drive when it comes to small random reads/writes.

My workload is a lot more dependent on random IO than most, and I *still* can't really say with complete honesty that the price difference of the top-end MLC drives is worth it (though SLC is another story...)

 

people are often not aware on how many random small reads happen even when you access just one file sequencially. parsing the file system and interpreting what to start to open the file with and all is tons of small random spread reads (sysinternal tools can show you that).

same for writes, of course.

there's no such thing in an os as a single sequencial access, when EVER you do something on disk, it's a bunch (up to 100ts) of random accesses. that's just the way it goes.