New Indilinx firmwares coming with up to 500% performance?

Toggle sidebar Toggle sidebar
Engineer

Engineer

Elite Member
Oct 9, 1999
39,230
701
126
Click me...

Saw this at OCZ and thought it was interesting. Have read that even some of the generation one Vertex drives might get this as well as other manufacturers. Looks to be NAND specific.

Here's hoping that some out there will get a big boost! :)
 
Zap

Zap

Elite Member
Oct 13, 1999
22,377
7
81
The question is are they only going to release it for past Indilinx powered OCZ drives, or for any brand that used those controllers? I ask this because... didn't OCZ acquire Indilinx? Would be a nice boost for my Patriot TorqX drives.
 
frostedflakes

frostedflakes

Diamond Member
Mar 1, 2005
7,925
1
81
This is really interesting. I'm kind of interested in how much write amplification will be reduced.
 
E

Emulex

Diamond Member
Jan 28, 2001
9,759
1
71
why don't we just hack the firmware to reduce life to 1 year ish and then reflash it back to normal and send it in for warranty. alot of people that "CHIP" their cars do this. turbo up the boost from 12psi to 19psi then flash it back before service. very common practice.

Anyone know how to decode the firmware? I bet we could make the sandforce and indilinx controllers go mad on speed by disabling most of the life saving techniques. I'd be glad to have 2-3x the speed for 6 months life honestly. 10 years is what they last now. 6 months-1 year would be great. especially if you used double your credit card warranty dealio on a 1 year warranty drive.
 
N

nanaki333

Diamond Member
Sep 14, 2002
3,772
13
81
Emulex said:
why don't we just hack the firmware to reduce life to 1 year ish and then reflash it back to normal and send it in for warranty. alot of people that "CHIP" their cars do this. turbo up the boost from 12psi to 19psi then flash it back before service. very common practice.
.
Click to expand...

except now most stealerships can tell if/when you chip or flash your car, and how many times you've done it. only way to really get around it is replace your whole ECU.

if enough people hack their ssd firmware, manufacturer's are bound to void warranties like cpu & video card manufacturers void warranties of OC'd devices.
 
E

Emulex

Diamond Member
Jan 28, 2001
9,759
1
71
people overclock cpu's and are their warranties void? video cards? nope. I think it's time to overclock storage!
 
Zap

Zap

Elite Member
Oct 13, 1999
22,377
7
81
boochi said:
OCZ forum staff have stated that competitors will get the firmware before OCZ users will.
Click to expand...

That won't necessarily mean we'll see them first elsewhere, because some of the other companies may want to validate or customize (branding) the firmware.
 
V

Voo

Golden Member
Feb 27, 2009
1,684
0
76
Emulex said:
people overclock cpu's and are their warranties void? video cards? nope. I think it's time to overclock storage!
Click to expand...
Well apart from the fact that overclocking a CPU/GPU will indeed void your warranty ;) Vendors are just not bothering in most cases because it's hard to establish and only done by a small minority anyhow, so it's cheaper to write the losses of I wager.

And I doubt your proposal of hacking the firmware in such a way is possible. Disabling some function calls, changing return values and so on - sure that's easy. But pretty much changing the inner workings of a complex code with only the assembly (and then that's no detailed documented asm like for x86,..) available seems pretty impossible.
 
Last edited:
J

jiffylube1024

Diamond Member
Feb 17, 2002
7,430
0
71
I have two Indilinx SSD's -- I'll believe a 300% increase when I see it. True, Indilinx is about 300-400% behind on random read speed compared to Intel, Sandforce 1xxx series, but we'll see if/when this firmware comes out.
 
B

BigDH01

Golden Member
Jul 8, 2005
1,631
88
91
RU482 said:
supertalent just sent an e-mail regarding an update from v2030 to 2141.
the Yikes part was that it mentioned a bug in the old firmware:

http://www.supertalent.com/home/forum/viewtopic.php?f=74&t=10867

I would assume this applies to any Indilinx drive. Not sure what OCZ firmware corresponds to supertalent's 2030.
Click to expand...

Makes me nervous too. I own an early model Vertex 120 and the SMART data is giving me some pretty high erase count numbers. It also doesn't help that somewhere along the line one of the FW updates updated SMART to reflect the fact that the flash in my drive is only rated at 5000 writes, not 10000 (which was the number given to Anand in their review).

The official response on the OCZ forum when someone gets past the rated erase count is to tell the person to do a destructive flash. They have had some drives have panic failures past the erase limit, but a destructive flash has brought them back (I consider this to still be a pretty major bug). They also claim that SMART data is wrong... but it's got to be recording something although nobody seems to know what it really means. All in all, I'm not sure I trust OCZ right now or have much faith in them delivering on this FW.
 
bulanula

bulanula

Member
Apr 20, 2011
76
0
0
BigDH01 said:
They also claim that SMART data is wrong... but it's got to be recording something although nobody seems to know what it really means. All in all, I'm not sure I trust OCZ right now or have much faith in them delivering on this FW.
Click to expand...

Typical OCZ behaviour : "we don't know what the SMART data means" and yet you claim to have produced these SSDs LOL.

My rule of thumb for storage and RAM and panels and everything etc. : only buy from people that make the actual things eg Intel make the NAND but OCZ do not. LG make the panels but Digimate don't. Crucial make the ram but Mushkin don't etc. Best to get products from the manufacturer directly rather than go to a rebranded product.
 
V

Voo

Golden Member
Feb 27, 2009
1,684
0
76
bulanula said:
My rule of thumb for storage and RAM and panels and everything etc. : only buy from people that make the actual things eg Intel make the NAND but OCZ do not. LG make the panels but Digimate don't. Crucial make the ram but Mushkin don't etc. Best to get products from the manufacturer directly rather than go to a rebranded product.
Click to expand...
You do understand that IMFT is actually a different company from Intel and that the process of talking to the guys creating the flash is most certainly extremely similar for both camps? (It's not as if Intel was a small unbureaucratic company - heck those two groups are most certainly in different states)

And I'm still not sure why the hell we always end up with the flash. Because so far, nobody - nobody - was able to show me even one instance of a SF or Intel drive that ran out of flash cycles. So all the problems come down to the controller itself - well no wonder, because after all Intel and OCZ are using exactly the same flash.
 
Last edited:
G

groberts101

Golden Member
Mar 17, 2011
1,390
0
0
furthermore.. anyone that thinks that OCZ actually "makes" these drives doesn't understand this industry very well. Most all but the largest companies simply outsource the parts/builds while having the SSD controller chip mfgr write the code needed to get it working. If it needs additional changes?.. it goes back with a revision submission.

And now that Intel and Sandforce are lip-locked together(told ya so), they will surely have the clout and R&D necessary to alleviate many of the known Sandforce controllers compatibility issues when their versions hit the market soon.

Is probably a major part of the reason that Intel is dragging its feet right now to fix all it's drivers LPM bugs that so many P67/Z68 mobo users are dealing with when using the new V3. They pound on the door and Intel just says.. "just a minute.. I have to get dressed first". lol

Add all the hot swapping and power mgmt insufficiencies and it's easy to misunderstand the true cause of the current Sandforce clusterfluck many have to deal with right now.

And to stay on topic a bit here. Why wouldn't a company that just bought the Indilinx controller push for a new improved firmware to boost sales? They've already made the Turbo firmware available to older 50nm drive owners and this one will be available after the others adopt it first. Is actually a smart move if you ask me as it promotes their products future growth without being so greedy to take advantage of it themselves before any others can get a leg up on them. Intel sure as hell wouldn't be so giving.. that's for sure.
 
B

bradley

Diamond Member
Jan 9, 2000
3,671
2
81
frostedflakes said:
This is really interesting. I'm kind of interested in how much write amplification will be reduced.
Click to expand...

This very same question was asked on OCZ's forums, I could only assume the question deliberately wasn't answered. Very few questions regarding Indilinx's, comparatively speaking, horridly unacceptable write amplification numbers have been allowed on their official forum.

OCZ's main concern is literally having thousands of SSDs returned within the warranty period, including mine. I won't be buying OCZ products to be possibly fooled a second time; at least with my next SSD, I'm sticking with Intel.
 
G

groberts101

Golden Member
Mar 17, 2011
1,390
0
0
you seem to be confusing panic locked drives(which by the way can be recovered almost 100% of the time without RMA with a user requested destructive flash) with one's that have expended all the write cycles.

I have yet to see 1 single Indilinx drive that has suffered from burnt nand. Higher WA or not, most don't have a clue as to how much these things can take. Especially the older 50nm drives.
 
B

bradley

Diamond Member
Jan 9, 2000
3,671
2
81
groberts101 said:
you seem to be confusing panic locked drives(which by the way can be recovered almost 100% of the time without RMA with a user requested destructive flash) with one's that have expended all the write cycles.

I have yet to see 1 single Indilinx drive that has suffered from burnt nand. Higher WA or not, most don't have a clue as to how much these things can take. Especially the older 50nm drives.
Click to expand...

I could always say that you are confusing (to obnoxious redundancy) this forum with OCZ's. :) However, I do agree that current 34nm MLC lifespan is likely higher than current estimates.

Otherwise, there is no confusion relating extremely high write amplification (estimated at 10x+) with many times quicker read-only status. Indilinx's controllers have done an extremely poor job at keeping write amplification low, and instead are many times higher than the competition. Whether my drive is rendered useless prior to the 3-yr warranty period is anyone's guess.
 
Last edited:
G

groberts101

Golden Member
Mar 17, 2011
1,390
0
0
"obnoxious redundancy" is rampant around this place and I shouldn't be criticized for adding to it any more than you should be.

and a more accurate generalized statement would be.. "I agree that ALL MLC lifespan is likely higher than current estimates" as we are no closer to burning these things out then we were years ago when we moved every write possble off these drives.

Fact of the matter is that whether you have 10,000PE/c nand with high WA or 3,000PE/c nand with lower WA?.. the numbers often crunch to around the same lifespan in the end. Firmware has improved. GC has improved. Static data rotation has been implemented by many which all on its own can result in much more effective wear leveling. And then there's the basic fact that capacity alone is the largest contributor to life expectancy of any SSD regardless of PE/c or WA.

So, constantly bitching and complaining about past issues with any hardware or business ethics to "warn others"?.. is no better than traveling around to try and add perspective with an over-protective fan-boy attitude. lol

And just to be clear here?.. I know the limitations and shortcomings of OCZ very well, just as anyone else well versed in this particular industry but need not go into any details as there are more than enough around here that are willing to smear it all out on the table. However misinformed or reliant on speculation they may be to form an opinion about it. Then add all the "I got burnt" guys?.. and the picture is pretty easy to see.

I've got 15 of their drives in this house(with well over 40 different SSD tested so far) and I just "might" have a better perspective then some who have cryed foul after having issues with just 2.

You are right on target about one thing though. If you are not the tweaker type who doesn't mind finding workarounds or using the occassional destructive flash?.. stick to an Intel for fewer issues. Errr.. wait.. now that Intel is in bed with Sandforce?.. they will finally get the bugs worked out and ride Microsoft and the mobo mfgrs to apply fixes. It will be enjoyable to see all the hipocrites show up to use the new Sandforce based Intel drives once they hit the shelves later on. We're all for sale to the highest bidder and I'll take speed over reliability any day. :)
 
E

Emulex

Diamond Member
Jan 28, 2001
9,759
1
71
so who'se in on hacking the firmware to ditch all wear leveling and kick the iops into high gear? who cares if it lasts 3 years a 60gb ssd will be nothing in 3 years. peanuts. and you can always put the old firmware back on and sell it after you drive it like you sold it.

kinda like most people selling used video cards/cpu's here. they never post how fast they got it on air :) or water cooling. lol.

Let's do it! screw lifecycle i want speed!
 
G

groberts101

Golden Member
Mar 17, 2011
1,390
0
0
I'd be game but I know some others have already tried without success. I'll also try to find the link to the guy who recently pulled his apart over on the OCZ forums and had the balls to ask OCZ for directions. lol

he was monkeying around with baud rates and all sorts of stuff from the rx/tx pins on the edge of the PCB and thought he could make a go of it. Never saw him again after a few failures. There were a few around trying it when the controller was first released the year before as well.

The thing to remember is that there is misconception that all throttling is strictly there for improved lifespan. When in fact, as I've eluded to many times over there, the simple fact is that the worst case throttle(hammered state throttle or 50% max incompressible write speed under fresh state) is a simple limitation of the Sandforce recycling engines max performance capability.

IOW, there's some maketing spin being pushed about Durawrite whereas in fact that's as fast as the controller can read-modify-write to keep up. The newest controller has a larger engine and has faster on the fly cleaning with greater TRIM utilization capability so it can do better. But limitations are still there and is not possible to ever have these things maintain full speeds 100% of the time. Maybe some day, but certainly not yet.

Errr.. NM.. I just realized we're talkin' about Indilinx here. There's that "too much multitasking/breezing through" thing again. lol
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom