New Intel 520 Benchmarks

[=Wendy=]

I've never seen 35 on a 4k read on any drive. Thats definitely impressive. I get 25 on my 830.

Most of the larger capacity SF-2281 based SSDs with sync NAND will manage around that speed. The Vertex 3 for example will manage 33.97 MB/s (source) with CDM.

 

[jwilliams4200]

I've never seen 35 on a 4k read on any drive. Thats definitely impressive. I get 25 on my 830.

That 35MB/s from CDM for 4KiB random read is seen for many Sandforce 2281 SSDs, but it is a bogus number. If you run AS-SSD or IOMeter (set to full random), the true number is observed, and it is around 25MB/s. The higher number from CDM is an artifact of how CDM measures.

 

[=Wendy=]

That 35MB/s from CDM for 4KiB random read is seen for many Sandforce 2281 SSDs, but it is a bogus number. If you run AS-SSD or IOMeter (set to full random), the true number is observed, and it is around 25MB/s. The higher number from CDM is an artifact of how CDM measures.

I'm not saying CDM reports it correctly, but Anvil's Storage Utilities reports very similar results to CDM for 4K random read, with compressible and incompressible data.

So, it's not that easy to just say CDM is wrong.

 

[bryanW1995]

I'm not saying he is wrong but its very hard to gauge the "feel" of it. I don't think he is a newb at all but it's difficult for others since they can't feel this and benchmarks don't show a 2x improvement. There may be other things are play here and yes I know it was a drive image so there shouldn't really be any differences when it comes to the OS.

I would love to get my hands on one of these drives but I will be staying on my SATA2 board for awhile longer plus i'm adding a second intel 160 G2 so I can go Raid 0 when the 11.5 RST are released.

Those ssd's don't degrade very much even without TRIM, and if you somehow got into a bind it's already possible to download the beta 11.5 RST, anyway.

BTW, I read about the g2's on Anand's original writeup about them, but I've also experienced it firsthand with my x25m g2 80gb in raid 0 (SRT cache + data) for the past couple of months. Or, to put it better, I HAVEN'T experienced the need for trim on my g2 in past couple of months.

 

[Obsoleet]

I've never seen 35 on a 4k read on any drive. Thats definitely impressive. I get 25 on my 830.

Although if you look at the two drives on bench, the 830 actually wins more tests than it loses (without using pen and paper) and many of the scores are very close so I do find it hard to believe there is a 2 fold in noticeable speed.

Of course there's not. Even with Intel's validation, the only thing you might see 2-fold is BSOD reports. This is Sandforce (the shadiest of the modern controllers) with Intel (the most reputable of modern manufacturers), and I'm not convinced Intel has powered through the dungpile that is Sandforce yet. Definitely going to be interesting in the next 6-12 months.

 

[Coup27]

Of course there's not. Even with Intel's validation, the only thing you might see 2-fold is BSOD reports. This is Sandforce (the shadiest of the modern controllers) with Intel (the most reputable of modern manufacturers), and I'm not convinced Intel has powered through the dungpile that is Sandforce yet. Definitely going to be interesting in the next 6-12 months.

Schoolboy error on my part for forgetting about CDM being compressible.

We've another Sandforce horror story going on in the "SSD keeps disappearing thread".http://forums.anandtech.com/showthread.php?t=2225438

 

[Makaveli]

Those ssd's don't degrade very much even without TRIM, and if you somehow got into a bind it's already possible to download the beta 11.5 RST, anyway.

BTW, I read about the g2's on Anand's original writeup about them, but I've also experienced it firsthand with my x25m g2 80gb in raid 0 (SRT cache + data) for the past couple of months. Or, to put it better, I HAVEN'T experienced the need for trim on my g2 in past couple of months.

Ah interesting with that being said I may install the second ssd sooner its just sitting on my desk.

I didn't know a beta was available.

 

[bryanW1995]

It will be extremely interesting to see how this pans out, though something tells me that intel just would have killed the project if they couldn't get it up to their standards. Having said that, I'll let others beta test it for me... 🙂

Ah interesting with that being said I may install the second ssd sooner its just sitting on my desk.

I didn't know a beta was available.

I almost installed it, but I hadn't noticed any degradation yet, anyway. Realistically, by the time you think you want it (if ever), the final driver version will be out. Plus, even if you never trimmed those 2 drives in RAID 0, they'd still be faster than one x25m that ran at new specs forever.

 

[slow_poke]

That 35MB/s from CDM for 4KiB random read is seen for many Sandforce 2281 SSDs, but it is a bogus number. If you run AS-SSD or IOMeter (set to full random), the true number is observed, and it is around 25MB/s. The higher number from CDM is an artifact of how CDM measures.

All the tests show different numbers for same part of a test. If you want to put more emphasis on one test over another that would be up to you.

As SSD will always be a worst case scenario for any SF drive, atto the best case, and all the rest somewhere in between, so take your pick.

Two 520's in R0 for over two weeks now...

 

[palladium]

...how did you get 300MB/s for 4k random write?

 

[slow_poke]

No matter what bench the 4k writes always come out at around the same 300 mb/s. The above test was with random data, here is one with 0-fill 😀

 

[groberts101]

...how did you get 300MB/s for 4k random write?

Intel raid caching benefits being added.

 

[palladium]

Intel raid caching benefits being added.

Makes sense

 

[Elganja]

ugh... i hope the prices of the 256gb one becomes more reasonable...

looking to replace all my SSD's I currently have in my main PC (2 40gb g2's and 2 160GB g2's)

 

[imusorka]

Does this look right to you all? New Intel 520 180GB.

 

[BFG10K]

Necro thread - Super Moderator BFG10K.