ASUS RAIDR Review...finallly!

[ARod]

I'll pass on this until it's finally done right, not that ASUS did a bad job with the tools on hand.  Just glad someone was finally able to do a proper review on it.

 

http://us.hardware.info/reviews/4263/asus-raidr-express-240gb-pci-express-ssd-review-is-this-the-future

 

[MzCatieB]

i do hope this isnt going to be expensive, i realy want to use one in my next build just for the sake of it

[ARod]

i do hope this isnt going to be expensive, i realy want to use one in my next build just for the sake of it

 

If you're looking for an SSD then it makes sense.  If you already have an SSD then just pass on it.  In order for it to be viable it can't be much more than the leading edge SSDs.  Otherwise, there's no point in purchasing it.

[MzCatieB]

If you're looking for an SSD then it makes sense.  If you already have an SSD then just pass on it.  In order for it to be viable it can't be much more than the leading edge SSDs.  Otherwise, there's no point in purchasing it.

it means less wires in the case, neater look for the build im doing

[ARod]

it means less wires in the case, neater look for the build im doing

 

True, and makes sense if you're looking for an SSD.  But, if you already have an SSD it's just blowing money for the heck of it or for enthusiasts with OCD tastes ;)

 

The price will be the true decider, wish they would release that.

[MzCatieB]

True, and makes sense if you're looking for an SSD.  But, if you already have an SSD it's just blowing money for the heck of it or for enthusiasts with OCD tastes ;)

 

The price will be the true decider, wish they would release that.

yh the only bad thing about it is the red... i will have to respray it from red to white :3

[eshbop]

TBH if I had the option of PCIe or a SATA SSD, assuming similar price (that won't happen), performance (well it should be faster) and everything runs smoothly, I'd go for the PCIe. Too bad these are using Sandforce controllers though.

[ARod]

yh the only bad thing about it is the red... i will have to respray it from red to white :3

 

lulz....use your lowest PCI-e slot, you won't see the red.

[Nicktrance]

 

More importantly, the ASUS RAIDR Express 240 GB will cost about twice as much as a Samsung 840 Pro 256 GB. If you really need the high sequential transfer rates, you could buy two 840 Pros and put those in RAID 0. You're still much better off then in price per gigabytes. 

 

You have to be suffering from an extreme case of OCD or too rich to care to use these right now, especially considering the performance is not really that much better.

[MzCatieB]

lulz....use your lowest PCI-e slot, you won't see the red.

i will know O.O

i might actualy just change the cover on it

[mgsstar]

Would match a GD65 gaming board! :)

[Eric1024]

Blah, I hate to be a Debbie Downer, but this thing is too small (capacity), not fast enough to make it attractive to users, and definitely two expensive.

They could have fit so much more flash on that board if they wanted to.



Just look. Compared to a lot of pcie cards this pcb is damn near vacant.
The back is much worse. Very little going on here.

Granted this (below) is a raid controller and not an ssd, but look at the difference in component density:

Just look at how dense the components are. This is an lsi 9260 btw.

That and they're only using 16GB dies. Really guys? That controller can support up to 16x32GB dies, or 8x64GB dies for a total of 512GB of space.
Proof:
LSI's controller info: http://www.lsi.com/products/storagecomponents/Pages/SF-2200-2100.aspx
Other drives using that controller, may with up to 16 nand dies: http://thessdreview.com/tag/sf-2281/
Specifically: http://thessdreview.com/our-reviews/intel-ssd-525-multi-capacity-ssd-review-msata-ssds-just-became-mainstream/

Some of the intel 525 msata drives in that last link are using the SF2281 with 64GB nand dies, so it's definitely possible. So even if asus didn't want to try to increase component density and use 12x or 16x32GB dies per controller, they could easily just swap in 64GB dies and bam, you have a 1TB ssd. Then, they could make the whole board smaller and save the end user a *bit* of money.

Now that I've complained enough about design, let's discuss performance. This thing is ... well, slow. For about the same price as the 240GB version, I can throw together two 256GB 840 pro's or ocz vectors in raid 0. Those will give you double the capacity, slightly better the performance, as well as lower power consumption and better serviceability (you can replace one drive if one drive dies. On the RAIDR, if one of the two drives dies, you need an entirely new card. (Yes, you would loose all data either way but with raid 0 I would just assume backups)

And I have a few final thought/questions/answers. What's the market for this? Performance enthusiasts and power users. What do those two clientele groups look for in products? Generally, price, performance, and sometimes aesthetics. (As a side note, I will admit, this thing is absolutely gorgeous, no question about it).

One last thing... PCIe 2.0, really guys?

As with other high data rate serial transmission protocols, the clock is embedded in the signal. At the physical level, PCI Express 2.0 utilizes the 8b/10b encoding scheme^[27] to ensure that strings of consecutive ones or consecutive zeros are limited in length. This coding was used to prevent the receiver from losing track of where the bit edges are. In this coding scheme every eight (uncoded) payload bits of data are replaced with 10 (encoded) bits of transmit data, causing a 20% overhead in the electrical bandwidth. To improve the available bandwidth, PCI Express version 3.0 employs 128b/130b encoding instead: similar but with much lower overhead.

http://en.wikipedia.org/wiki/PCI_Express#Data_transmission

Okay so time for some more math before I explain what that all means. PCIe 2.0 spec is 500MB/s/lane (that's megabytes, not bits), which means the 2x interface on the RADR gets 2x500MB/s or 1GB/s of total bandwidth. BUT WAIT. PCIe 2.0 has 8b/10b encoding, meaning that only 8 bits of actual data get sent for every 10 bits transmitted (the rest is used for overhead and a kind of signal error correction and prevention, if you will).
This means that the 1GB/s that the device would otherwise have is really only 800MB/s. This explains why the drive never broke 800MB/s in the sequential tests (and makes me wonder how the hell ASUS thinks they got over 800MB/s in any test as per their specs).

However, as evidenced in the ATTO QD4 1MB write benchmark on page 6 of the review that op posted, the drive is very much hitting that celling.

It seems to me like they were going for something extreme but kind of cut off half way there. What could they have done? (here comes the fun hypothetical part where we get to imagine things and drool)

1. Easy, so easy, put 64GB or 32GB dies on it and make a 512GB/1TB model (for the love of god people will buy it, that's one of the things PCIe ssd's stand out for: capacity.

2. I know this is a bit harder to quantify and a bit stupid to suggest because it sure as hell isn't this simple, but they really should have tried to target this thing for a lower price point. If they had a 512GB version for 650$, I'd strongly consider it.

3. Granted the drive doesn't perform anywhere close to the 800MB/s pcie 2.0 2x limit except in one test, they should still at least allow the drive some bloody breathing room. Give it either PCIe 2.0 4x or PCIe 3.0 2x ... please? I mean really. That seems like such a basic early design decision that they just blatantly ignored.

4. Okay, this one's a little out of reach, but this is obviously an extreme product (or trying to be), so why not make it actually extreme? Why not put 4 ssd controllers on it in raid 0 (or better yet, 5) so the thing can really smoke sata drives in sequential speeds. OR just spend some time optimizing the bloody firmware so the thing can actually perform better as it (if you don't believe firmware optimization can do much, go read about intel's DC s3700 ssd. Link http://thessdreview.com/our-reviews/s3700/

I'm not going to lie. I was very excited for this product, but as it stands I'm very disappointed. I hope with a few generational iterations ASUS can actually become a major player in the ssd market, because generally they make very good products. (though as it stands the ssd market is dominated by he who has the fab (intel, micron/crucial, and samsung), so that's probably not going to happen)

[Borgodoc]

.

[Sotti]

Remember, this is their first attempt at a PCI-E SSD. It obviously has it flaws, but let's hope they learn, improve and give us a satisfying product in the end. I trust Asus enough to believe that they eventually will release some baller product.

 

Yes but definitely I think this RAIDR is not that one.

 

I've read that the 480Gb model could be better cause it would have 4x120Gb in RAID0. Let's see the price point.

 

Read you

[James_AJ]

From what I'm seeing, for the estimated price point, it's better to just SATA III up two S3500's in RAID 0...

faster, cheaper, longer durability ermahgerd

[Exotath]

Pretty cool to have a new competitor in the SSD market. Performance isn't where it should be but if the price is right I would buy it.