Post your AS SSD scores

[Rubycon]

8X 256GB SEC 830 RAID0 4K stripe Areca ARC1880ix-24 host.

ASSD is kind of dated on newer drives and smart hosts. Anvil works better as you can select a work size up to 32GB so caching does not play a big role in work numbers. It's IOMeter based.

 

[Emulex]

http://sudomakeinstall.com/linux-sy...-lsi-9266-8i-controller-in-a-dell-r720-16-bay dang someone trumped me.

well I'll try anyways.

 

[reynoldsjrmy]

Hi,

2 x OCZ Vertex 4 256GB in R0:

Regds, JR

 

[fffblackmage]

I don't have anything blazing fast, but I recently did some benchmarks between my brother's Crucial M4 128GB, Samsung 830 128GB, and OCZ Solid 2 60GB, and my own 2x OCZ Agility 60GB and Vertex 2 EX 55GB. Old stuff, but still, they should serve as decent comparisons to see how far we've come in the SSD world.

Unfortunately, these aren't all done with the same computer, but I'll at least note what chipset the SSDs are attached to.

First up, the M4 and 830 (running on a Z68-based rig):

OCZ Solid 2 60GB (F/W 1.5, on Z68 rig):

OCZ Vertex 2 EX 55GB (F/W 1.37, E-350 Zacate/A55 Hudson-based rig):

My Agility 60GB (F/W 1.7, AMD 785G-based rig):

And my other Agility (F/W 1.7, PM965-based laptop). The Intel chipset seems to do a bit better than my desktop's AMD chipset.

After upgrading both Agility's to Arowana FTL. Same as the previous two - the former is on my 785G-based rig and the latter is on the PM965-based laptop:

Before I updated the Agility's to Arowana FTL (basically, they're still on 1.7), I ran both Agility's in a RAID0 array in my 785G-based rig, 64k stripe size:

Disappointed by the poor performance, I installed them into the Z68-based rig to set up another RAID0 array with 64k stripe size:

Hopefully, that wasn't too confusing. The conclusion I can see is that Intel just seems to do a bit better than AMD, especially when it comes to RAID arrays (shocking, right?).

 

[ThePazza]

So I recently built a new computer and contrary to multiple people's advise I created a RAID0 using 4 x Samsung Evo 840 120GB drives. I think the results speak for themselves in giving the skeptics something to chew on. :-D

 

[baron10]

what about this

 

[Fernando 1]

what about this

Or this:

 

[Z15CAM]

4 GB 1866 Mhz 9-9-9-24-1T @ 1.35v RAMDisk Vs 256MB Samsung 840 Pro

I should mention that my Boot SSD is mounted as a None Array Member on the SATA Raid Controller with 2 X's 500GB WD Blacks in Raid-0 and a single 1TB WD Black for Backup and storage mounted on the RAID Controller as a NONE Member of an Array.

 

[BonzaiDuck]

These are about what I'd expect, given the Magician and CrystalDiskMark results. However, since the system has been running for days, I'm wondering if the RAM available to RAPID is somehow less and different than if I ran the test immediately after boot-time:

 

[ctk1981]

My dads computer doesn't have AS-SSD but I thought this amusing considering he has an old school WD 750GB HD, sata II.

Yeah Rapid on samsung drives can make it look great, but there is other commercial software out there that can do the same thing. His system is an old C2D 2.1Ghz, 4Gb DDR2 800mhz etc etc..in other words "ancient".

Edit: For those curious, a measely 1.4Gb of ram is dedicated to the cache vs the new Rapid 1.1 that uses up to 4GB.

 

[tidy]

.

 

[BonzaiDuck]

. . . .
Yeah Rapid on samsung drives can make it look great, but there is other commercial software out there that can do the same thing. . . .

This is worthy of a comment. Definitely -- I agree.

Especially on a machine using anything other than the Samsung SSDs, your only real choice is "other software." I think there were a few options cited, and I remember Romex Software's PrimoCache. Better or worse, I wouldn't know.

And as I may have said somewhere already, the technology and the semi-proprietary nature of some hardware devices makes it more challenging for software-houses to universally address their chosen objective or problem solution.

Take for instance Alfredo Comparetti's "SpeedFan." It must accommodate several different types of sensor devices and motherboards. With SpeedFan, the prodigious list of motherboards folded into revisions of the software accompanies a lag or delay in producing those revisions, and people still find SpeedFan daunting or confusing.

It may be likely that certain brands of 60GB SSDs work better than others in the Intel ISRT caching configurations. If that is true, then again -- there is the uncertainty that a "one-size-fits-all" software and firmware solution can't catch every bug, glitch or incompatibility. The best reason that a firm like Intel would put forward such a technology derives from their perception of being a major SSD and chipset producer.

Not so with the strategy of using system RAM for caching. DDR3 is . . . well . . . DDR3. Flour is flour, corn is corn, sugar is sugar -- and only fools will try and tell you that Betty Crocker or Del Monte or C&H is any different from the generic grocery store brand.

So here comes Samsung. They would obviously create a proprietary barrier to the application of the software. Why would they enable (and empower) rival SSD makers with something that otherwise makes the Samsung SSD line stick out from the crowd? With software design, that's their prerogative, and they probably know well of the other SW houses like Romex or the potential of those sorts of caching strategies.

So now, ctk1981 has me thinking. I've got computers with Intel or Crucial SSDs -- perhaps I should get the Romex or similar software. But when I look at ctk1981's results, I have to temper that thought.

He's cached a WD drive. The caching software gives him sequential reads over 1,500 MB/s. But the writes are consistent with the HDD. Even so, you would suspect that only so much cache can be had from 4GB of RAM, whether you cache an SSD or an HDD. Conversely then, the Samsung results with my 16GB RAM is totally consistent with ctk1981's results, or consistent within a ballpark.

Howsoever it shakes out for general or proprietary caching software, proof of the pudding is in a history of stability and reliability with whatever hardware is used. That's easier, because RAM is just a commodity for how it works among several motherboards and chipsets.

 

[ctk1981]

This is worthy of a comment. Definitely -- I agree.

Especially on a machine using anything other than the Samsung SSDs, your only real choice is "other software." I think there were a few options cited, and I remember Romex Software's PrimoCache. Better or worse, I wouldn't know.

And as I may have said somewhere already, the technology and the semi-proprietary nature of some hardware devices makes it more challenging for software-houses to universally address their chosen objective or problem solution.

Take for instance Alfredo Comparetti's "SpeedFan." It must accommodate several different types of sensor devices and motherboards. With SpeedFan, the prodigious list of motherboards folded into revisions of the software accompanies a lag or delay in producing those revisions, and people still find SpeedFan daunting or confusing.

It may be likely that certain brands of 60GB SSDs work better than others in the Intel ISRT caching configurations. If that is true, then again -- there is the uncertainty that a "one-size-fits-all" software and firmware solution can't catch every bug, glitch or incompatibility. The best reason that a firm like Intel would put forward such a technology derives from their perception of being a major SSD and chipset producer.

Not so with the strategy of using system RAM for caching. DDR3 is . . . well . . . DDR3. Flour is flour, corn is corn, sugar is sugar -- and only fools will try and tell you that Betty Crocker or Del Monte or C&H is any different from the generic grocery store brand.

So here comes Samsung. They would obviously create a proprietary barrier to the application of the software. Why would they enable (and empower) rival SSD makers with something that otherwise makes the Samsung SSD line stick out from the crowd? With software design, that's their prerogative, and they probably know well of the other SW houses like Romex or the potential of those sorts of caching strategies.

So now, ctk1981 has me thinking. I've got computers with Intel or Crucial SSDs -- perhaps I should get the Romex or similar software. But when I look at ctk1981's results, I have to temper that thought.

He's cached a WD drive. The caching software gives him sequential reads over 1,500 MB/s. But the writes are consistent with the HDD. Even so, you would suspect that only so much cache can be had from 4GB of RAM, whether you cache an SSD or an HDD. Conversely then, the Samsung results with my 16GB RAM is totally consistent with ctk1981's results, or consistent within a ballpark.

Howsoever it shakes out for general or proprietary caching software, proof of the pudding is in a history of stability and reliability with whatever hardware is used. That's easier, because RAM is just a commodity for how it works among several motherboards and chipsets.

With regards to the write performance....on that particular benchmark run, I did not enable Lazy write in supercache v5 (or deferred write as it is on primocache) which does boost write scores on SSD's. I'll do a quick bench on his system with lazy write enabled to see how much of an increase.

To explain the difference between the two benchmarks. Original on the left was done with defualt page file size which is 16K and no lazy writes. New bench on the right is done with pagefile size of 256K which reduces overhead and increases speed, lazy write enabled at default 5 second value. Still a 1.4GB ramcache.

 

[Z15CAM]

Too many Synthetic Benching in this thread.

 

[tidy]

.