• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

SSD degredation / TRIM vs GC

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Whats holding him back is the ADFD drive hes got paired up with the Velo.
I knew that. (see my edit an hour before your post). :)

Yeah SSD degredation...its without a doubt there when it gets filled up. They running TRIM or GC?

Also, I found quite an interesting post on the OCZ forums regarding GC and TRIM from Tony:
The issue is peoples perception was TRIM would keep the drives at full speed, the fact is it does not due to data fragmentation and the lack of block consolidation. So then you add drive level GC which can mess up the way win7 looks after how it deals out TRIM commands as GC alters what the volume bitmap should look like.

The only draw back with GC is added writes, you can massively add life to a drive by over provisioning the Nand..EG you build a 60GB drive with 60GB spare for wear leveling but this costs a lot of money. So if you use GC just be mindful what you use the drive for..(data storage on regular HDD etc)


Adding a little more, win7 still is the route of all evil really for SSD, what is needed is a steadystate cache system within win7 that traps all writes, sorts them into sequential full block fills and rewrites them in low IO periods, this way the OS is read only, writes are sequential not random and TRIM would have a much easier time.

I sent mine in to Ryder to flash to the GC FW, and since my computer never sleeps/idles as I [email protected], the writes issue should be non existent. This also will allow me to upgrade to the TRIM FW if Im not liking how the GC works out as you cannot flash backwards.


(sorry about the minor threadjack Joe)
 

Ashura

Member
Joined
Oct 4, 2003
Location
NYC
Win7 is the "route of all evil" for SSD?

I'm sorry, but while I think some of the OCZ SSD products are great and OCZ Support can be great, I'm so sick of their people constantly ****ting on the OS as the source of their SSD hardware and firmware problems. Can an OS be completely optimized to be perfect for SSDs? Absolutely. But those optimizations don't exist now and you're selling to a market that has no choice but to deal with whatever OS is out there today. Your product should be tailored to work in that market without any problems and if it doesn't, you should fix the hardware issue on your end instead of constantly blaming the software.

They dumped on the OS when the Core drives came out, even though MTron, Samsung and Intel SSDs were stutter free.

When Intel's G1's serious degradation issue was discovered, was blame placed on the OS? No. Intel took notice of the problem and released a firmware issue to resolve it.

Sorry for the rant but as someone who was an early adapter of the JMicron-based OCZ Core series that basically got ****ed with bad hardware that should've been fully tested before being sold, that kind of talk infuriates me.
 

Joeteck

Retired
Joined
Oct 5, 2001
Location
Long Island
Win7 is the "route of all evil" for SSD?

I'm sorry, but while I think some of the OCZ SSD products are great and OCZ Support can be great, I'm so sick of their people constantly ****ting on the OS as the source of their SSD hardware and firmware problems. Can an OS be completely optimized to be perfect for SSDs? Absolutely. But those optimizations don't exist now and you're selling to a market that has no choice but to deal with whatever OS is out there today. Your product should be tailored to work in that market without any problems and if it doesn't, you should fix the hardware issue on your end instead of constantly blaming the software.

They dumped on the OS when the Core drives came out, even though MTron, Samsung and Intel SSDs were stutter free.

When Intel's G1's serious degradation issue was discovered, was blame placed on the OS? No. Intel took notice of the problem and released a firmware issue to resolve it.

Sorry for the rant but as someone who was an early adapter of the JMicron-based OCZ Core series that basically got ****ed with bad hardware that should've been fully tested before being sold, that kind of talk infuriates me.

Thank you for a well thought out reply / RANT. Was well worth the read... And I can feel your frustration... I'm looking forward to your article submission.
 
OP
EarthDog

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
I liked the reply too, one thing to note though is that post was taken out of context. I emplore Ashura to read a couple posts around that one and understand the context that statement was written in.

What I believe he (Tony) may be saying is a standard method at the OS level would benefit ALL SSD's and their performance. I do not think its a blame game.
 

Ashura

Member
Joined
Oct 4, 2003
Location
NYC
Thank you for a well thought out reply / RANT. Was well worth the read... And I can feel your frustration... I'm looking forward to your article submission.

Thanks, I'm hoping it'll be worthwhile and useful. :)

What I believe he (Tony) may be saying is a standard method at the OS level would benefit ALL SSD's and their performance. I do not think its a blame game.

That's exactly what he's saying, by having an OS where all writes are cached and then written at once in large blocks you take advantage of an SSDs high seq. write speed and aren't dependent on random write speeds. This was the only true solution to OCZ's (and other manufacturer's) JMicron drives, currently only executable through Windows SteadyState. I tried it a year ago with the OCZ Core drives:

Decided to go ahead and try the solution out for myself. I've been using an OCZ Core V1 64gb w/ SteadyState for a week now and it works great. Stuttering is completely gone and the drive is very snappy/fast with regards to apps, extracting files, etc. (I'm comparing this to my previous experience with a 2x36gb Raptors in RAID0 as an OS drive, and a 300gb VelociRaptor as an OS drive, the OCZ Core with SteadyState feels faster than both). The two biggest negatives are 1. the long boot time 2. that you can't copy and move around files bigger than your cache (as explained in the OP's link) and 3. reduced disk space thanks to the cache. 1 and 3 are annoying but bearable as I don't shut down that much and I have an external drive where I keep most of my storage. 2 is fine as I set the cache to 10gb and I don't plan on moving anything bigger than that around.

It's, as of today, a poor solution and only good if you're willing to compromise for the negatives like I was. And, it's primary purpose is to make up for low random write speeds. If you've already got blazing fast random write speeds like the Intel G2 drives have even with TRIM enabled over time, then you won't get much of a benefit from this.

I know about this because it's nothing new from OCZ, and that's exactly my problem. A year ago when OCZ older generation SSDs were out, they were saying the exact same thing and dreaming about the exact same OS solution. The problem is you shouldn't be designing today's products for a non-existant OS of the future and you shouldn't be blaming today's OS for your design flaws.

To OCZ's credit, they do design their SSDs properly now and they do take into consideration input from the enthusiast community. When the Vertex was first coming out, they were going to release a drive with lower random write speeds and higher seq. speeds. But Tony created a poll in the OCZ forums and found that most users want the exact opposite, for good reason, and they've been creating SSDs with that mindset ever since. It's why I own a Vertex in addition to my X25-M and why I'll be wholeheartedly recommending them in my article.

But I still hate little digs at the OS by OCZ like the one you quoted because I think it's a ridiculous position to hold. I'm probably overreacting. :beer:

As far as GC vs TRIM goes, I took Tony's quote to mean that you should be running either GC or TRIM, not both at the same time.

And this is also MAJORLY off topic. Might not be a bad idea for a mod to move all primarily SSD focused posts to another thread like 'Guide to SSD Selection'. ;)
 
Last edited:
OP
EarthDog

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Great call on the moving these posts. I will report my post, the head of this monster and make that request since I doubt any mod will just waltz on by and see your statement. :)

Yes, that is what Tony is saying regarding GC and TRIM, but the point of my posting that info was to say that TRIM, as noted by Tony, "may not be" the best answer as far as keeping performance up on a drive. It was my question in that thread he was replying to. :)
 

Ashura

Member
Joined
Oct 4, 2003
Location
NYC
Yes, that is what Tony is saying regarding GC and TRIM, but the point of my posting that info was to say that TRIM, as noted by Tony, "may not be" the best answer as far as keeping performance up on a drive. It was my question in that thread he was replying to. :)

I took it to mean, unless you have a drive that doesn't support TRIM or RAID, that TRIM is the best answer. Remember that we should be discounting any comparisons to a dream-SteadyState-type OS as an alternative. If you fall into either of the two above categories, then GC is the best answer. But using both at the same time is not a good option. I could be interpreting that completely wrong though, I'll be catching up on OCZ forums' massive threads on the topic though to verify.
 

tuskenraider

Senior Member
Joined
Apr 23, 2002
Right... also keep in mind that TRIM is only a function on W7. So GC would have to be used on every other situation...right?
You also need to use GC/GC based firmware if you run SSD's in RAID, TRIM will not work even in Win7.
 
OP
EarthDog

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Tony and Perfectdisk10 FTMFW!!! Good find!

I wonder if this can work with TRIM or GC?
 
Last edited:

Ashura

Member
Joined
Oct 4, 2003
Location
NYC
I wonder what the best way to test effectiveness of the different methods would be. Perhaps perform the exact same writes on three different occasions with my Vertex, once with TRIM firmware, once with GC firmware and once with stock firmware and PD10 and compare speed afterwards?
 
OP
EarthDog

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
I just sent it in to Ryder for the GC firmware flash...so TECHINCALLY no. It was apersonal handoff.
 

Ashura

Member
Joined
Oct 4, 2003
Location
NYC
I would just wait till a proper working TRIM firmware is released, it shouldn't be too long from now. Your drive will be fine till then.

But yes, if you didn't want to wait, then it's a viable alternative.
 

Automiketic

Member
Joined
Jun 21, 2008
well the only reason i dont use trim is because by my understanding it is not supported in vista?