Haswell looks impressive when using SSDs in RAID

[MasterC Program]

RWLabs just posted a great article about the RAID performance of Haswell. It looks pretty good and seeing this I might upgrade after all

http://www.rwlabs.com/article.php?id=836

 

[SithSolo1]

Wouldn't that have more to do with improved motherboard chipsets than Haswell?

 

[mavere]

The Z87's controller has arbitrary, unexplained performance drops that the Z77 didn't, and it's supposed to be better?

We've come a long way since the terrible days of JMicron; let's not regress again. Modern SSDs are not supposed to be a game of roulette.

Edit:

Just to clarify, Z87:

Z77:

 

[MasterC Program]

Is that a wide spread issue ? Couldn't it just be SSD degradation ?

 

[24HZ]

Whats the point of SSD raid with top shelf SSDs? You're not going to improve random access times, only sequential speed. What good is going from 500Mb/s to 1000Mb/s? Will anyone even notice that?

 

[CakeMonster]

The Z87's controller has arbitrary, unexplained performance drops that the Z77 didn't, and it's supposed to be better?

We've come a long way since the terrible days of JMicron; let's not regress again. Modern SSDs are not supposed to be a game of roulette.

Edit:

Just to clarify, Z87:

That's no clarification. Posting an image link with no explanation of the conditions of testing, no hardware/driver info, nothing at all.. is just FUD. If you have an issue to report, give us some proper info.

 

[willomz]

Even if there is an issue there will be a BIOS fix before long anyway.

 

[nanaki333]

Whats the point of SSD raid with top shelf SSDs? You're not going to improve random access times, only sequential speed. What good is going from 500Mb/s to 1000Mb/s? Will anyone even notice that?

IOPS go up too. servers use raid SSDs as well.

wow! look at that increased 4k performance! wonder what intel did to squeeze that out.

 

[JoeRambo]

There is something shady going on with 4k RAID0 performance. For example on my system with freshly secure erased drives and fresh windows 7 install, I am getting just 26MB random read out of 2x256GB 840PRO ???? While sequential read is 1GB+, 4k read performance is disastrous. I really hope Anand will investigate and put Intel's corporate rear end on fire for such a product in 2013.

Edit: setting cpu in windows power settings to minimal 100% (cpu pegged @ 3.5Ghz all test) improved 4k random read performance - now getting 37MB/s, but still real low...

 

[mavere]

That's no clarification. Posting an image link with no explanation of the conditions of testing, no hardware/driver info, nothing at all.. is just FUD. If you have an issue to report, give us some proper info.

Of course, people reading the article? What a terrible idea!

Sheesh

 

[Vesku]

Of course, people reading the article? What a terrible idea!

Sheesh

I read the article, how is it at all informative without running the same SSDs on a Z77 Ivybridge system and comparing?

 

[mavere]

I read the article, how is it at all informative without running the same SSDs on a Z77 Ivybridge system and comparing?

I'm not entirely sure what you mean, but the necessary Z77 and 840 Pros comparison points were linked on top of each applicable page.

 

[IntelUser2000]

IOPS go up too. servers use raid SSDs as well.

wow! look at that increased 4k performance! wonder what intel did to squeeze that out.

I wonder if the behavior has to do with Dynamic Storage Accelerator feature in Z87. Cooaler result was looking pretty similar to this one.

 

[Vesku]

I'm not entirely sure what you mean, but the necessary Z77 and 840 Pros comparison points were linked on top of each applicable page.

What I mean is why don't they directly compare them in their review, they link to a previous review of the same drive using a 2500K (presumably stock vs overclocked Haswell, why is it overclocked for a storage review? Does it effect any of the tests vs stock?) and Z77. The previous Z77 review uses a different version of the Intel RST driver. The only difference I could tell from the single drive (the Z77 review they link to doesn't test multiple drives) was 4K (writes) performance and the difference is not present at a Queue Depth of 32 at least in Crystalmark (I'm not wading through the entirety of both reviews), shouldn't they have run the test on a Z77 platform with latest storage driver especially if they are going to talk about how impressive the Z87 platform is??

Why as the reader am I having to navigate multiple z77 reviews to get an idea of the differences? Just looked at their 2 drive raid and followed the Z77 link, 4K with Queue Depth is better for Z77. At least for the 2 drive raid they are using the same driver version.