LSI Acquires Sandforce...

[e-drood]

as described by http://www.tomshardware.com/news/lsi-sandforce-ssd-acquisition-controller,13836.html & ANAND & "others"...

is it possible that the ocz purchase of indilinx contributed to this / perhaps intels interest in sandforce also served to arrange this marriage (however prettied up lsi's take-over of sf is)

sf is beginning to learn the realities of cohabiting with the real "owners" of the sandbox --- intel samsung marvel come to mind

and you really think lsi will make any effort to pick-up sf controller liability presently outstanding...
the fly-on-the-wall in tmsc's conference room must be hearing an earful this morning...

time to unload those ocz warrants or shares or did you simply receive a sweet promise whispered in your ear...

 

[lsv]

Good I say, LSI already makes great controllers. Maybe they'll start making LSI Raid + Sandforce controlled SSDs in the near future?

 

[Shmee]

interesting, I hope we get some great new innovation from this.

 

[zlejedi]

Hmm why does the opp sound unhappy with the news?
LSI= great storage controlers, years of experience in enterprise bussiness
Sand Force= producer of controllers plague by serious bugs, advertised by fake* transfer speeds

This probably won't help people who have SF 2xxx series but maybe 3k or 4k series Sand Force will be worth using

*achieved only in best case scenario of data compression

 

[nanaki333]

Good I say, LSI already makes great controllers. Maybe they'll start making LSI Raid + Sandforce controlled SSDs in the near future?

great controllers? are you sure you're talking about the same LSI?

they have ok controllers, but i wouldn't call them great. i used to buy LSI for low end servers when i didn't have a big budget and they were a great deal for low cost hardware raid.

 

[lsv]

great controllers? are you sure you're talking about the same LSI?

they have ok controllers, but i wouldn't call them great. i used to buy LSI for low end servers when i didn't have a big budget and they were a great deal for low cost hardware raid.

I should have written 'for the home user looking to set up back up personal solutions'. They are no Adaptec or Highpoint that's for sure but I think LSI purchasing Sandforce will work out for the better. Hopefully they integrate LSI raid with Sandforce controlled drives and make something interesting.

 

[nanaki333]

I should have written 'for the home user looking to set up back up personal solutions'. They are no Adaptec or Highpoint that's for sure but I think LSI purchasing Sandforce will work out for the better. Hopefully they integrate LSI raid with Sandforce controlled drives and make something interesting.

yeah. that's MUCH better. for low cost real hardware raid, they're great.

i'm hoping to see something nice out of this union. hopefully they don't sit on it like ocz did after they bought out indilinx.

 

[e-drood]

this link presents concise factual history of LSI

http://en.wikipedia.org/wiki/LSI_Corporation

this acquisition will not result in synergy or increased market share or more reliable/more advanced controller design - expediency drives this rush to the alter...

read lsi's many faceted history...

there are certainly better summaries of LSI, but this is simpler to link

no, not angry - perhaps disappointed

 

[e-drood]

re: sandforce "product marketing":

sf's controller business is split (2) ways - high volume mech hdd controller chips & lower volume nand controller chips used in ssd's

as i've written, it's a disappointing marriage...

 

[zlejedi]

no, not angry - perhaps disappointed

Still even if they kill Sand Force totally that only would benefit market.

Sand Force is perfect example of marketing taking over engineering.

This controller offers companies ability to advertise impresive transfer speed thanks to it's compression algorithms but there's a lot of ugly truth hidden behind this once you dig deeper.

And existence of drives with asynchro memory cells is even worse for the market since something like Agility 3 appears to be almost as fast as Vertex 3 (on the box) for much lower prices.

Even Intel and Kingston were forced to join the sand force party since much better competing controllers can't match the "on the box" speeds of SF.

So whatever LSI do with SandForce it can only be improvement over current situation.

 

[e-drood]

well said

raising 1st round venture funding (in western countries) is driving "marketing over (valid) engineering"

and this is unlikely to change near term

 

[nanaki333]

hopefully lsi has more talented firmware writers than SF too.

 

[groberts101]

I'd like to know the reasoning behind the thought process of "SF compression is all just smoke and mirrors". Obviously all the reviews around the net aren't just based on the boxes advertised specs and ARE beating out other controllers on merit alone.

If not then do you not think that "our saviour's" light and heavy test benchs have no relevancy to ACTUAL usage and overall performance levels?

and back to the topic.. if it goes through?.. then you will in fact see some decent PCI-E raided solutions. Modular PCI-E designs which you can build(add to) as you go would be killer.

 

[e-drood]

"on merit alone" --- product merit equals transparent plug-n-play WITHOUT challenging user in-depth tech intervention - ie: NO Voodoo rites to install...

the "benches" DO NOT reflect actual user experience due to realworld data flow patterns --- "benches" are marketing hype at best...

sf compression smoke-n-mirrors --- speaking of marketing hype - read the sf/founder assigned "patent" based business model as originally presented to 1st round venture fund subscribers - aggressive compression algorithm as patented sought to overcome silicon controller limitations WITHOUT adding additional "arm's" (which samsung has now done / 3 replacing 2) - ultimately this aggressive compression model introduces additional issues in controller stability while complicating the user installation experience --- THE MARKET PLACE IS NOW SPEAKING BY WIDELY DISCOUNTING SF 2000 SERIES SSD DRIVES --- THE MARKET PLACE DETERMINES "PRODUCT MERIT" NOT "BENCHES"...

"if it goes through" --- it is as already accomplished - additional round of venture funding has been withdrawn due to critical issues with BOTH sf 1XXX series AND 2XXX series nand controllers...

LSI will attempt to "cherry pick" from SF's limited intellectual properties store to apply to raid products - but SF in-house simply can not write functional firmware and has now paid the price for this by losing control of their company...

Sandforce had exhausted it's in-house boutique expertise and would shortly join the "uncounted multitude" of failed tech start-ups...

LSI has been picking at rubbish tips for two+ decades & has managed to survive in it's niche market(s) --- this speaks volumes about the real state of SF's hughly hyped original business model everyone bought into...

 

[frostedflakes]

Even in real-world scenarios SandForce drives perform very well. They're great controllers, the biggest problem is the compatibility issues and other firmware bugs they've been plagued with. That's the reason people have been steering clear and choosing Intel, Crucial, and Samsung drives instead.

 

[groberts101]

Even in real-world scenarios SandForce drives perform very well. They're great controllers, the biggest problem is the compatibility issues and other firmware bugs they've been plagued with. That's the reason people have been steering clear and choosing Intel, Crucial, and Samsung drives instead.

Thank you!.. Sheesh I didn't even have the patience to read thorugh all that abstract and redundant drivel he keeps posting around here.

Dude acts like Sandforce killed his whole family or something. :'(

 

[e-drood]

i tip my hat to one of the ocz masters of positive spin

however this does not change the reality of the continuing sf controller problems

but its good to read you have kept ur sense of humour - at the expense of others

 

[=Wendy=]

Thank you!.. Sheesh I didn't even have the patience to read thorugh all that abstract and redundant drivel he keeps posting around here.

Dude acts like Sandforce killed his whole family or something. :'(

Just add e-drood to your "ignore list". It works for me. :thumbsup:

 

[Coup27]

I had difficulty reading it simply because of how it was written.

 

[HendrixFan]

the "benches" DO NOT reflect actual user experience due to realworld data flow patterns --- "benches" are marketing hype at best...

Sorry, you are absolutely incorrect. Starting out with an easily disprovable false premise like that really makes anyone question the rest of your unsupported claims.

http://www.anandtech.com/show/4902/intel-ssd-710-200gb-review/9

http://www.anandtech.com/show/4902/intel-ssd-710-200gb-review/10

http://www.anandtech.com/show/4863/the-samsung-ssd-830-review/4

http://www.anandtech.com/show/4863/the-samsung-ssd-830-review/5

Anand's two newest SSD reviews from a month ago shows the real world data flow patterns show the usefulness of the Sandforce drive. And lest you think the test is biased towards compressibility see what Anand says in the first link:

Only 42% of all operations are sequential, the rest range from pseudo to fully random (with most falling in the pseudo-random category). Average queue depth is 4.625 IOs, with 59% of operations taking place in an IO queue of 1.

Take the time to actually read up on what Anand has done for his heavy and light storage benches so that you can appreciate his reviews and the results he is able to show us.

 

[e-drood]

come sit by the hearthside and share good feelings and be of good cheer

hendrixfan, the anandtech benches are the creation of anand & staff and exclusively represent anandtech (public) website take on storage performance parameters - within the framework of editorial independence compromised by advertising revenue support --- this is not a subscription based website which does not accept paid advertising...

further, said "anand benches" do not reflect any widely recognized standards of storage performance - those benches are entirely a personal take on the storage world...

do the benches in question provide a relative product comparison - yes, in a limited sense - the test hardware platform(s) employed represent an ideal hardware collection and/or current ideal hardware collection - while NOT representing the global reality of at least (2) generations of hardware components/elements assembled together into a pc/workstation

this has significant impact on the validity of ANY prevailing storage benchmarks

benches are infotainment used by manufacturers & resellers as marketing hype... perhaps this is not the intention (of benchmark authors) but it is the reality

a relative measure, yes; an accurate measure, no... the concept of a universal test hardware platform is flawed and invalid and insulting to users...
******

ok u need the illustrated version, the ragpicker is deconstructing sf as we speak & ur much needed future firmware updates are up-the-chimney
******

good friends, good cheer, fond memories

 

[HendrixFan]

hendrixfan, the anandtech benches are the creation of anand & staff and exclusively represent anandtech (public) website take on storage performance parameters

I am well aware of that, that is why I brought it up. In the link Anand explains that he felt synthetic benchmarks weren't representative of what users would do and as such they wouldn't give good enough information on the worthiness of storage products.

further, said "anand benches" do not reflect any widely recognized standards of storage performance - those benches are entirely a personal take on the storage world...

Now you are just playing both sides of the argument. First you said the "benches DO NOT reflect actual user experience due to realworld data flow patterns--- "benches" are marketing hype at best..." which I would mostly agree with. But the Anand storage benches are different. They do reflect actual user experience and real world data flow patterns. They don't follow a "recognized standard", because that is exactly what synthetic benchmarks are.

Please take the time to learn what the storage bench is and how it works.

 

[e-drood]

hendrixfan, this is the crux of the issue:



do the benches in question provide a relative product comparison - yes, in a limited sense - the test hardware platform(s) employed represent an ideal hardware collection and/or current ideal hardware collection - while NOT representing the global reality of at least (2) generations of hardware components/elements assembled together into a pc/workstation


the actual user hardware platform determines the "realworld" user experience with either a hdd/nand hybrid or ssd; the large majority of global users do not enjoy an ideal (or even current) hardware platform

a more representative "test" hardware platform(s) would be in the interests of all users all regions

 

[HendrixFan]

You want the part being tested to be the limiting factor. It is senseless to test out 40 different computer configurations when dealing with SSDs, because unless you are running a P4 with 1GB of RAM you won't be limited in seeing a performance boost from a SSD.

Now that we have SATA 6Gbs drives, there is a difference when running a newer SSD on a SATA2 channel. Those are benched because they actually limit performance.

You can set up tons of configurations but they will give you the same results. Computers are bottlenecked by storage speeds. Not the other way around. If other components would bottleneck a SSD then a new configuration would matter. Aside from the new SATA standard, there is nothing in a modern computer that would bottleneck a SSD.

The Anand storage bench gives accurate and fair results, different systems would not show much of any variance from them. The only real variance you see is in the SSD drives, which is what the benchmarks show.

 

[VirtualLarry]

why, oh why, isn't there an "Add to Ignore List" option when you click on someone's username.