A note from Intel...

[sub.mesa]

Actually... background GC is really good. I can watch my Indilinx drives (have both background GC and TRIM) stay basically flat in terms of performance over time in Raid 0. Current array was built about two months ago and it is running 2-3% under new. It really changes your perspective on the need for TRIM in RAID.

Yes but keep in mind Indilinx is entirely different and also has much lower random I/O scores to begin with.

It is true that Intel seems to be the most affected by performance degradation over time. One argument could be the limited spare space that is used by default, only 6.8% while Sandforce by default has a massive 28% as spare space (i.e. 100GB advertized but 128GiB physical NAND). But also the level of intelligence employed by the controller would have a primary role here. For example, simple flash devices like compactflash and USB stick have no spare space or performance degradation at all; they are just always slow when doing random writes and are serial operation based (no NCQ/TCQ).

Indilinx is a simple yet effective controller that sustains good writes particularly, but is rather weak when looking at both random reads and writes; though single queue random reads are still as good as Intel; as here it isn't handicapped by its four-channels, while Intel has ten. On Windows this isn't such a bad disk really, but on a server UNIX system the high-queue depth would make Intel SSDs fly while the Indilinx would fall behind considerably.

Sandforce and Micron appear to be real competitors to Intel, but they are still perfecting their firmware. It's not clear what Intel will be introducing this year in November/December, but i still consider them to have the upper hand in technology; only their controller is truely native with onboard SRAM buffer; as far as i know other controllers don't have this and have to use the slower DRAM for every internal operation. This would also explain Intel's very low latencies both read and write; nobody competes with those.

But Intel's sequential write is quite bad; most apparent on the 5-channel 'capped' Intel X25-V; it pushes only 40MB/s; half of it's X25-M 80GB brother. But it's the random reads and writes that Intel is quite good at. If the 5-channel X25-V already does 180MB/s+ then that ought to mean the X25-M series is interface/controller-capped.

 

[Emulex]

i see 230-250mb/s reads from the x25-v. i think like automakers - they underrate the important aspects. that many smb/home users don't do alot of writing so 90-95% of the time their cheapest product will perfect at 80-100% of their most expensive product. of course if everyone knew this - they would choose less margin products right? that would be good for nobody.

 

[DrPizza]

Scott, looking at your credentials (majored in Literature?!), are you just a public relations/marketing via blog type of guy, or do you have other credentials which enable you to accurately answer technical questions?

 

[postmortemIA]

i see 230-250mb/s reads from the x25-v. i think like automakers - they underrate the important aspects. that many smb/home users don't do alot of writing so 90-95% of the time their cheapest product will perfect at 80-100% of their most expensive product. of course if everyone knew this - they would choose less margin products right? that would be good for nobody.

I see >200MB/s reads on Windows XP with few years old tools, but fresh tools such as Crystal Mark show about 190MB/s.. still better than advertised 170 MB/s

 

[pjkenned]

Yes but keep in mind Indilinx is entirely different and also has much lower random I/O scores to begin with.

It is true that Intel seems to be the most affected by performance degradation over time. One argument could be the limited spare space that is used by default, only 6.8% while Sandforce by default has a massive 28% as spare space (i.e. 100GB advertized but 128GiB physical NAND). But also the level of intelligence employed by the controller would have a primary role here. For example, simple flash devices like compactflash and USB stick have no spare space or performance degradation at all; they are just always slow when doing random writes and are serial operation based (no NCQ/TCQ).

Indilinx is a simple yet effective controller that sustains good writes particularly, but is rather weak when looking at both random reads and writes; though single queue random reads are still as good as Intel; as here it isn't handicapped by its four-channels, while Intel has ten. On Windows this isn't such a bad disk really, but on a server UNIX system the high-queue depth would make Intel SSDs fly while the Indilinx would fall behind considerably.

Sandforce and Micron appear to be real competitors to Intel, but they are still perfecting their firmware. It's not clear what Intel will be introducing this year in November/December, but i still consider them to have the upper hand in technology; only their controller is truely native with onboard SRAM buffer; as far as i know other controllers don't have this and have to use the slower DRAM for every internal operation. This would also explain Intel's very low latencies both read and write; nobody competes with those.

But Intel's sequential write is quite bad; most apparent on the 5-channel 'capped' Intel X25-V; it pushes only 40MB/s; half of it's X25-M 80GB brother. But it's the random reads and writes that Intel is quite good at. If the 5-channel X25-V already does 180MB/s+ then that ought to mean the X25-M series is interface/controller-capped.

I have a feeling you know I run Indilinx, Sandforce and Intel G1/G2/ X25-V's also But totally agreed on Indilinx being a decent enough Windows drive. Realistically, I can't tell them all apart except if I am looking at benchmark results.

Another thing is that the Sandforce drives are now starting to get cheaper than Intel on a cost/GB basis. $200 for a X25-M 80GB v. $250 that I paid for my Agility 2 120GB a few weeks ago. With SandForce basically at cost/ capacity parity with Indilinx drives, Indilinx either needs to drop prices or there is no real reason to look at their drives at this point.

 

[kalniel]

By the way, is anyway else attending the summit? I heard there might be a surprise appearance!!

Postville refresh (25nm)?

I'm interested to hear where you're going with that one - performance must be hitting against Sata 2 limits on client setups, so do you support Sata 6gbps? But are Intel willing to do that when there are only third-party controllers, or is it going to tie up with new Intel chipsets and whatever controllers they may have onboard?

 

[Edrick]

Postville refresh (25nm)?

I'm interested to hear where you're going with that one - performance must be hitting against Sata 2 limits on client setups, so do you support Sata 6gbps? But are Intel willing to do that when there are only third-party controllers, or is it going to tie up with new Intel chipsets and whatever controllers they may have onboard?

I am interested in this as well.

 

[Old Hippie]

performance must be hitting against Sata 2 limits on client setups, so do you support Sata 6gbps?

Maybe they'll jump to Light Peak?

 

[JSt0rm]

By the way, is anyway else attending the summit? I heard there might be a surprise appearance!!

Lord Solid State and the Flashmems!!! they are my favorite swooooon :wub:

 

[Zargon]

haha, i hope so. This is me:
http://blogs.intel.com/technology/authors#scott_dwyer

i-l-l

 

[mutz]

Scott, looking at your credentials (majored in Literature?!), are you just a public relations/marketing via blog type of guy, or do you have other credentials which enable you to accurately answer technical questions?

that's exactly the issue..., they send this guy here which probably understands less then some of the users here, that's they're approach, keep treating people like clueless children and they are all very respected people, with ties and suits,
totally BS.

can't figure how people are eating this crap and shut up, this is arrogance btw if you have asked sub.mesa.

 

[Zargon]

that's exactly the issue..., they send this guy here which probably understands less then some of the users here, that's they're approach, keep treating people like clueless children and they are all very respected people, with ties and suits,
totally BS.

can't figure how people are eating this crap and shut up, this is arrogance btw if you have asked sub.mesa.

his degree means nothing about his credentials.

I am a network admin with a bachelors in Psychology :sneaky:

it certainly doesnt mean he can't understand the stuff

 

[sub.mesa]

Let him come forward with his intentions and confirmation that he is here to answer technical questions, which is what he stated on the first page if you read correctly.

He also said he would answer my questions 'next week' if he's able since he also will be busy with the summit; so just be patient, please?

 

[mutz]

He also said he would answer my questions

yeah, what he can... (if you read correctly :hmm:..)

This is the kind of stuff I came here to answer (not that I can and will answer everything in your post)

that's what i'm saying zargon..

but whatever, i was saying these things for you to see and understand, not even against him or anything..

fine, let's put aside all mess with Intel and see what he brings.

 

[Old Hippie]

fine, let's put aside all mess with Intel and see what he brings.

I'm I the only one to notice that the "mess" follows you?

 

[coolVariable]

Scott - why no TRIM for us first gen intel SSD owners?

 

[Zargon]

yeah, what he can... (if you read correctly :hmm:..)

that's what i'm saying zargon..

but whatever, i was saying these things for you to see and understand, not even against him or anything..

fine, let's put aside all mess with Intel and see what he brings.

yup and I see your point

but I also think if you back up and take a deep breathe, reread the thread. you appear to be seriously jumping his shit for no apparant reason.

we just want you to give the guy a chance

 

[mutz]

but I also think if you back up and take a deep breathe, reread the thread. you appear to be seriously jumping his shit for no apparant reason.

i know what i'm saying and why i'm saying it, i would have taken the time explaining it all, what is wrong with it, what is arrogant and what is selfish in it and the answers,

basically people do not do things out of they're will to make things better,
mostly people are doing it out of interest, and where is interest, there is no love in what you'r doing,
basically you'r taking care for your own self and not the other, you soup them up with pretty words when all you want is your own benefit or rather they're money...
i went explaining/showing it here a few times and mostly got answers like "this is business, that's what happens in business, you obviously knows nothing about that, wake up.."

which means, who ever said that, simply don't understand basic things about life, life within society and the true meaning of them.

so i'm just being cautious, i've ended few discussions here just to stop arguing with people about things they just skip automatically,
there are many things to be said here, which are much more important then these answers within that blog, and require a much wider view of things, there are many things to notice, things that reveals others, you learn by the small things...

i kind of like to keep it so currently yet i'm open.

 

[Skiprudder]

Hey, lets not bring major into this. Brian Reynolds, the head of Big Huge Games (Rise of Nations, Catan) is a history and philosophy major. Majors are starting points, they don't define who we are.

 

[mutz]

Majors are starting points, they don't define who we are

?
elaborate..?

 

[Skiprudder]

I was just pointing out that just because our Intel friend here is a Lit major we shouldn't assume he doesn't know what he's talking about. If we all ended up doing at 40 exactly what we picked to major in at 18 or 19 I think the world would be a much duller place. He is at Intel instead of say Knopff, so I expect him to know something about technology.

?
elaborate..?

 

[mutz]

I was just pointing out that just because our Intel friend here is a Lit major we shouldn't assume he doesn't know what he's talking about.

it was not coming out of a prejudgment...

 

[MrDudeMan]

i know what i'm saying and why i'm saying it, i would have taken the time explaining it all, what is wrong with it, what is arrogant and what is selfish in it and the answers,

basically people do not do things out of they're will to make things better,
mostly people are doing it out of interest, and where is interest, there is no love in what you'r doing,
basically you'r taking care for your own self and not the other, you soup them up with pretty words when all you want is your own benefit or rather they're money...
i went explaining/showing it here a few times and mostly got answers like "this is business, that's what happens in business, you obviously knows nothing about that, wake up.."

which means, who ever said that, simply don't understand basic things about life, life within society and the true meaning of them.

so i'm just being cautious, i've ended few discussions here just to stop arguing with people about things they just skip automatically,
there are many things to be said here, which are much more important then these answers within that blog, and require a much wider view of things, there are many things to notice, things that reveals others, you learn by the small things...

i kind of like to keep it so currently yet i'm open.

I honestly have no idea what you are attempting to say. Scott signed up to provide answers where possible. You took that to mean he doesn't know what he is talking about or doesn't have sufficient technical knowledge. What he meant was he most likely is not allowed to answer some questions and may have only a limited understanding of certain aspects of the technical side of things, but by no means should you assume he doesn't know anything. So he isn't an engineer - big deal? I am analog circuit engineer at Intel and there are plenty of things I don't know that someone who isn't an engineer might understand. Give the guy a break.

 

[mutz]

but by no means should you assume he doesn't know anything.

i really havn't said that..

i don't wish to take this thread back over again, it just kind of going round and round...
not everyone who smiles at you wants your happiness, it's nothing against that guy,
it is just looking at things from a different angle.

 

[Evadman]

SSDelightful, your silence on my issue is deafening.