The new SSD Thread (Benchmarks, Brands, News and Advice)

StratCat: The Corsair 128, although it's Corsair's "new" drive, is definitely without a doubt the "old" Samsung controller. The newest Samsung controller that has everyone impressed gives double the read/write speed of the Corsair 128.

 

TY ashura.

That's what I thought, but wasn't sure.

Good to see the *winning* response come from a fellow (long-time) OCC forum member.

+rep

 

if you wanted to make that point, then you really write funny sentences to try to say that. and i know, i'm the funny-strange-phrase-writer that nobody understands.

but yes. if the ocz would be perfect, i would have only ocz now. if the mtrons would be as cheap as ocz, everyone would have them. same for the intels. etc.

biggest advantage over the cache-thing: it works in every system, no matter how old (possibly a ide-sata adapter needed).

i can't wait for the bomb from hp.. memristors have more than 10x the storage at same production cost like flash, and no write issues. if that works out, they're at the price level of hdds, and have none of the ocz-problems.

but currently, it's still a dream..

 

you can still manually set it if you don't trust it. but i wouldn't bother about high offsets, it's not like you really lose much storage the only thing that matters is that they're aligned, and every what-ever-big power-of-two-number is.

no clue, acronis _should_ not change the offset, but i'm not sure. depends on how exactly it does the imaging.

but if you do a full-disk-image (not a partition-image), it should work. but i have no free disks to test it out to currently.

 

Well I assumed my Aopen MiniPC only supported SATA-1, but maybe it actually supports SATA-2? If it does support SATA-2 then I'll definitely be patient for a little while longer and wait for something better than the OCZ solid that is under $100 now. I was thinking that since the solid series SSD would max out the SATA-1 speeds that it would be no use in waiting for something faster since I wouldn't see the speed gains anyway.

How do I ascertain if my PC supports SATA-1 or SATA-2?

It uses Intel 945GM + ICH7-DH Chipset.

 

Question re: Intel X25-M performance.

I have a pair of X25-M in RAID 0 on my desktop, running 64-bit Vista. Looking at ATTO benchmarks, the read/write speeds are much higher than a single X25-M at the upper end of the benchmark, but much SLOWER at the lower end of the benchmark. Is this normal? Does it have to do with the stripe size of the RAID (sorry, I'm not a RAID expert), or is this indicative of a problem?

I realize that a lot of benchmarks are suspect with respect to SSDs at the moment, but still...

ATTO bench from legitreviews.com:




ATTO bench from benchmarkreviews.com:




My ATTO bench:




Furthermore, my results seem utterly crappy compared to what this gentleman is reporting.



His system:

Asus P5Q-Pro
Q9450 @3.4
8GB DDR2
HD4850
2x X25-M RAID0 connected to ICH10R 128k Stripe.
Vista x64 SP1
No "Tweaks" other than turned off page file, disabled defrag. That's all.


My system:

Dell XPS 630i
Q9550 (running at stock speed, no overclocking)
8GB DDR2
HD4850x2 (Crossfire)
2x X25-M RAID0 using onboard RAID on Dell MB. 128k stripe (I think).
Vista x64 SP 1


What gives?

 

The two formats are forward and backwards compatible. You can use SATAI drives in SATAII machines or SATAII drives in SATAI machines. Max throughput will be limited to SATAI speeds in both cases, though (not that there are too many drives capable of filling even the SATAI bandwidth).

 

I'm aware of the backwards compatibility, but I thought many SSDs more than filled the SATA-1 bandwidth? In other words, the speed of the Intel SSDs won't be realised on a system limited by SATA-1...moreover, it looks like up and coming SSDs will easily surpass SATA-1 speeds in both read and write...

 

Eh, I think that it's really just this generation of drives (starting with the Titan) that have started saturating the SATA1 bandwidth (disregarding the X25-M). Pretty much all of the simple JMicron MLC drives top out at 150MB/s and the Samsung and Mtron SLC drives don't get more than 100MB/s either. Maybe you won't get to maximize the read potential of an Intel X25-M, but you can still maximize the write potential which is really where all the cheap drives are failing.

 

correct but even then speeds are mostly only that high in benchmarks with IDEAL scenarios. Look at the speeds of the drives for benchmarks that supposedly simulate real life use. They r not near as high in most cases.

 

@heavyharmonies

I'm not overly familiar with RAID or if what u r reporting is an actual problem with your hardware but i do know hardware RAID will show far better performnce than your onboard RAID. Hardware RAID has a dedicated processor and cache and isn't usually limited by the SATA interface. (someone correct me if i'm wrong)

 

heavyharmonies: I believe Intel RAID is known to be vastly superior to nVidia RAID, that might have something to do with the difference in scores.

 

I work with large uncompressed audio files so I think that my situation approximates the ideal. If I understand correctly, loading an uncompressed audio file would be able to maximize the sequential read speeds of the drive. Right now it takes 2-5 minutes to load files depending on how large it is. That could be shortened significantly with an SSD for me...

So to ask my question again - how do I tell if my PC has SATA-1 or SATA-2? If it has SATA-1 then I can buy pretty much any current drive and be maxing out the SATA-1 bandwidth for reads. If my PC has SATA-2 then it would be worth waiting a little longer for some of the new SSDs to come out.

 

The mobile variant of ICH7, as used in netbooks and most likely what's used in your miniPC, is limited to SATA I.

 

According to the AOpen site, your AOpen board has the the ICH7-M which is limited to SATA1 speeds. Apparently the desktop ICH7 version can do SATAII though

 

Thanks to both of you

As I understand it the max real-world sequential read or write speeds I can get with an SATA-I interface is around 120-130MB/s. This means that there are *lots* of current SSDs that will max out the interface speed on sequential reads (and close with sequential writes), including the el-cheapo OCZ solid. I don't see a reason to wait for newer and better drives for my purpose since the speed increases (aside from random writes) won't be utilised by my hardware.

The way I look at it, I could buy a 7,200rpm drive (despite Aopen claiming that I should not) but that would cost almost the same as buying the 60GB OCZ solid at the current price of $96. Yes, the HDD would have more storage. Either option would be loads faster than my current 5,400rpm drive.

What would jesus do?

 

Take the plunge into SSD's if you go the Jmicron (Solid) route definitely hit up the OCZ forums for tweaks. I think i'm waiting to see what happens with vertex...

 

Well, SLC drives are largely made made for enterprise applications, and enterprise hardware, as we know, is the stuff of the Gods. Thus, Jesus would go and buy a 16GB or 32GB Mtron and call it a day

 

I've just done some preliminary testing of my new Samsung 256GB SSD, and it definitely handles stuff better than the G.Skill Titan. With the Titan, I couldn't run my VM at all until after I installed FMT, but the Samsung is running the VM fine with no operating system tweaks or SSD enhancing software like FMT. Still, the VM on the G.Skill with FMT seems faster than with the Samsung alone. I am still gathering benchmark data. I'll probably end up going with the Samsung + FMT. Also, the Titan seems to have been causing a lot of the heat issue with the notebook (Dell Studio XPS 16), because with the Samsung it's seems to be much cooler, still warm on the processor/video card side, but cool on the HDD side. Why would an SSD produce so much heat? Maybe the FMT was causing it somehow? I'll be running the Samsung with FMT tomorrow in my final test phase.

 

The dual JMicron drives run hotter than other SSDs simply because it's two controllers and a RAID chip versus just one controller.

 

The Titan has two JMicron controllers and a RAID controller on board. Given that a simple JMicron drive is usually dumps as much heat as a 5400RPM drive, and that the RAID controller is arguably a more power-hungry part; you're basically cramming 2-3 5400RPM drives in the space of one. It's gonna heat up a bit?

EDIT: That's the second time Ashura has sniped my post today

 

I had my Wheaties this morning.

 

Would it cause more heat than my wimpy 120GB 5400RPM 2.5" drive?
How about a 500GB 7200RPM HDD?

 

I see it already answered by Commander Wolf so your post was useful haha.

 

No, it (assuming you're referring to the Samsung) would have equal or less heat than the 5400RPM and definitely less heat than the 7200RPM.

 

So the G.Skill would be much hotter than the 7200RPM or equal?

 

Maybe I should start eating breakfast in general

2-3 times as hot. In theory. Maybe a little less in practice.

 

Ouch, that doesn't sound too good for my already heated m1530!

 

Yeah, I wouldn't recommend the Titan for laptops at all unless the laptop's going to be sitting on your desk for most of the time and heat isn't an issue. It's great for a desktop though.

 

so ashura, i saw u sold an X25-m on ebay the other day. What made you decide to dump that and wait on the vertex?

i guess it could just be same name but i figure ssd...u...

 

Yep, that was me. The Vertex numbers and $/gb are too tempting to not try out plus the Intel's wear leveling algorithm makes one have to factor in how to maintain consistent performance. I was never dead set on keeping a particular SSD to begin with, rather I wanted to find the best not-ridiculously-expensive one that had good performance and keep that (in the past two weeks I've bought, benched, used consistently for a few days and sold the Titan, the Cosair 128 and the X25-M). I'll try out the Vertex and if it lives up to the hype, great, but if not I'll have the Summit in my eye (or buy the X25-M again at what will hopefully be a cheaper price by then).

Out of every SSD I've tried so far, the X25-M still shines with the best real life performance, regardless of the problems with the wear leveling.

 

haha you lose/make much $$$ doing that?

 

You gave up on the X25-M?

I've still got my sights set on the Intel despite the apparent steady state slump... though I'm not as enthusiastic as I was before I heard about the performance degradation. I think I might just stick with Samsung and Mtron SLC as the prices (specially for Samsung) seem to be dropping pretty quickly and you still don't hear of much problems from these drives.

 

Not too much, the Corsair 128 was the biggest loss since there wasn't much demand for it.

Heh, giving up sounds too harsh. Like I said it's still the best SSD I've tried. More like I wanted to finance getting the Vertex, which may or may not end up outperforming (in both performance and $/gb value) the X25-M, and selling the X25-M was the easiest route towards that.

 

Furious debate going on over at OCZ about what level IO performance is desired/needed: http://www.ocztechnologyforum.com/forum/showthread.php?t=51364

I think xck had one of the best points:

http://www.ocztechnologyforum.com/forum/showpost.php?p=343902&postcount=43

Another good point from hahahafr:

http://www.ocztechnologyforum.com/forum/showpost.php?p=343939&postcount=52

 

Man, the heat difference with the Titan/Apex vs the Samsung is huge. I imagine this will result in better power savings with the Samsung, too. Cool. G.Skill Titan 256GB for sale, BTW

 

I'm glad to see IO's winning over in OCZ

 

It only took them this long to realize that their crappy IO's are the problem. It seems a little simplistic that they can simply fix this with different firmware as they are suggesting. If that was the case then they should have been able to fix stuttering a long time ago.

 

vertex isn't a Jmicron drive... problem drives are...

 

I had no idea until I did a search on ebay. There is a really low demand for Corsair. Why is that?

Corsair is not marketing this SSD at all.

 

didn't u just answer ur own question? not to mention the next gen samsungs are already out with double the specs...

 

Exactly, SSDs are already a niche market aimed at users who want the most performance. While a low power, low heat, no stutter, high GB, good $/gb drive at the cost of fast seq. reads/writes is appealing to laptop users, most who are buying SSDs now in the Corsair's $/gb range are willing to shell out the extra cash and sacrifice the 40GB to get the Intel or an SLC drive.

Does that mean it's a bad drive? No. It has it's place for people who place priority on the advantages noted above as opposed to super high reads/writes. But there just aren't that many of those people.

 

Because people are infatuated with super-high sequential read and write speeds and not enough consideration is given to random writes and IO. That's why OCZ and the lot are still pumping out these JMicron drives with just enough IO capacity to eliminate stuttering, but nothing more.

Then again, the Corsair/Samsung doesn't have that great of a random writing spec either, but it's still better than that of the JMicron lot.

 

I also wonder if most people are unaware that this is basically a samsung drive. 128gb samsung mlc drives sell pretty well on ebay (300-400)

 

Probably the relatively low read/write specs.

 

http://www.pcper.com/article.php?aid=669&type=expert&pid=1

Any take a read at this? What do you guys think?

 

Read the thread?

 

Yeah, it was posted a couple (tens?) of pages ago. That's basically to what everyone is referring when they talk about performance degradation on the X25-M.

 

Regarding partition alignment, i am under the impression that the starting offset is dependant of how big the flash block is, and that depends on what's used in each drive? The OCZ recommends 128kb offset, but can that be extrapolated to other drives? I read intel drives can only erase in blocks of 512kb, so it's partition should start with that offset?
Anybody can confirm this? If this is the case, then we have to know what block size uses each drive no?

Also, i think it's important not just to set the offset of the drive but set the correct partition size. Since we need to align windows allocation units with the disk's tracks (and also possibly the flash blocks?), things get even more complicated. The smallest common multiple of 4096byte allocation units and 32256bytes tracks (63 sectors of 512bytes) is 8 tracks, so the partition must be multiples of 8 tracks. I haven't factored in the size of flash blocks because i still need confirmation of what size a flash block is, and if it varies among manufaturers.

Am i mistaken?

 

Some good news, anandtech will be publishing a SSD roundup which includes the Vertex next Wednesday. We'll finally be able to see whether it lives up to the hype.