G. Skill 256GB Titan SSD in my MacBook Pro... Thoughts, Benchmarks, Pics, & Ramblings

[Thistledown]

I think more interesting than the benchmarks are the reports of responsiveness of the SSD's. The benchmarks look good, but the raw speed is only twice as fast for something that is about $1.90/GB (SSD) as opposed to $0.20-0.30/GB (HD).

However, the latency I think is where the speed benefits really make the difference -- the instant start up of apps due to not having to wait for the data on the disk to spin under the r/w head.

I really want to buy one, but I'm just not sure if it's enough space for me until the 512GB models come out at the end of this year.

 

[bcaslis]

Aside from speed, a benefit to me for an SSD is the complete lack of vibration or noise. If you use it on your laptop in silent conditions alot this is a benefit. If you use it only in a noisy office this probably doesn't matter.

 

[MrZebra]

If you use it on your laptop in silent conditions alot this is a benefit.

I hear you; I will buy one of those SSD as the price settles down a bit.

 

[tibi08]

Can anyone tell me how this compares to the Intel X-25 Mainstream SSD? I heard this was significantly faster (although, currently smaller in capacity and more expensive).

I can't wait to whack a SSD in my MBP tho....

 

[spaceballl]

Do you have any sense for your system running temperature with the old drive vs. this one? I have an early 2008 MBP and it can get pretty hot at times, so I'd not want to do anything that increased its running temp. I had bought a 320G 7200rpm drive to swap for my 250G 5400rpm drive that it came with, but this now has me thinking... looks like a big performance boost.

I will try and get a better sense for system temperature / battery life. I'll repost after I spend some more time with the new machine. Just from using it over the past couple days though, I think temperature seems about the same. It still gets warm in the are near the back of the keyboard / speakers - neither palm rest area seems to get that warm... but like I said, i'll update w/ better results after a few more days.

 

[spaceballl]

Can anyone tell me how this compares to the Intel X-25 Mainstream SSD? I heard this was significantly faster (although, currently smaller in capacity and more expensive).

I can't wait to whack a SSD in my MBP tho....

If you read the reviews I link to in my original post, they do comparisons against a lot of other SSDs, including the Intel.

 

[tibi08]

If you read the reviews I link to in my original post, they do comparisons against a lot of other SSDs, including the Intel.

Thanks, just looked, it's very impressive. I'd be tempted to buy now, but the technology is progessing as a scarily fast rate!

 

[spaceballl]

Thanks, just looked, it's very impressive. I'd be tempted to buy now, but the technology is progessing as a scarily fast rate!

It sure is! I'm sure that in 6 months there will be a new 256GB that is even faster and more power efficient than mine at a lower price than I paid. That's what makes technology so exciting, though, and I had to pull the trigger eventually!

 

[Lyshen]

I posted this in the other thread but if anyone is considering getting the Intel X25-M series and plan to use bootcamp... I haven't read a single success story of Intel X25-M SSD + Bootcamp working together.

Just a fyi.

 

[pprior]

Ok, you guys have convinced me. i'm ordering one of these bad boys. I'm sure it will be 1/2 the cost in 6 months, but I remember back to my 15K RPM scsi days how responsive those drives were, even with slower transfer speeds than we have now. I gave those up because having 4 of them was like living with a noise making machine running constantly.

My mac pro wasn't in today, it will likely be until middle of next week, so hopefully this will get here and be ready to install.

Does anyone have a recommendation on a drive adaptor that would let me mount this drive into the drive sled on the mac pro for plug and play?

Edit: ordered the maxupgrades adaptor. Looks beefy and should work with any future SSD.

 

[ShortArc]

I am having problems with my TITAN 256GB not waking after sleep on my Unibody MBP 2.8GHz. Placing the MBP to sleep by closing the lid works fine but if I leave the lid open and let the drive/computer go to sleep I am forced to reboot the machine. I know Spaceballl and others have not experience this phenomenon. If anyone has any insights or suggestions please let me know. I did try resetting the SMC but no luck. Also reinstalled the OS. I did not re-partition and format the drive anew yet but I don't see how that would make any difference.

 

[nippyjun]

I am having problems with my TITAN 256GB not waking after sleep on my Unibody MBP 2.8GHz. Placing the MBP to sleep by closing the lid works fine but if I leave the lid open and let the drive/computer go to sleep I am forced to reboot the machine. I know Spaceballl and others have not experience this phenomenon. If anyone has any insights or suggestions please let me know. I did try resetting the SMC but no luck. Also reinstalled the OS. I did not re-partition and format the drive anew yet but I don't see how that would make any difference.

Maybe they have it set that the drives don't go to sleep. Maybe it's not really needed with SSD's.

 

[ShortArc]

Well I swapped the SSD with a Samsung HM500LI having the identical disk image. Sleep mode (including deep sleep) all work fine.
So there is definitely something going on between my MBP and the TITAN SSD. I also did some looking around and there are a bunch of folks who are just having wake problems, like the one I described, not running SSDs. I am clueless for now.

 

[spaceballl]

Well I swapped the SSD with a Samsung HM500LI having the identical disk image. Sleep mode (including deep sleep) all work fine.
So there is definitely something going on between my MBP and the TITAN SSD. I also did some looking around and there are a bunch of folks who are just having wake problems, like the one I described, not running SSDs. I am clueless for now.

ShortArc,
Very odd problem you're having. So FYI when I said my computer was going in and out of sleep just fine, it was... but that was in regard to closing the lid and having it wake up again.

However, I just tested it by putting it in and out of sleep mode a few times manually (no lid closing) and that seems to work just fine as well. Very interesting issue you're running into - maybe swap it for a new one before your 30 day warranty runs up?

 

[tubbymac]

The only thing worrying about that Titan drive is that it still uses that piece of junk JMicron controller except that it has two of them instead of one. There are complaints on other forums (Anandtech and other hardware sites) of stutter/freeze issues while using this drive.

Also benchmarking on an SSD is misleading when you first get the drive. All MLC drives allow you to do a single write operation to a SSD memory cell the first time it is used. When you first get a drive obviously all the cells are unused and you will get super fast benchmarks. On EVERY subsequent write to a cell that has been written to before all MLC ssd drives require a read operation first, followed by either a write or an erase + write operation. This will significantly reduce performance and benchmarks.

 

[dgdosen]

The only thing worrying about that Titan drive is that it still uses that piece of junk JMicron controller except that it has two of them instead of one. There are complaints on other forums (Anandtech and other hardware sites) of stutter/freeze issues while using this drive.

Also benchmarking on an SSD is misleading when you first get the drive. All MLC drives allow you to do a single write operation to a SSD memory cell the first time it is used. When you first get a drive obviously all the cells are unused and you will get super fast benchmarks. On EVERY subsequent write to a cell that has been written to before all MLC ssd drives require a read operation first, followed by either a write or an erase + write operation. This will significantly reduce performance and benchmarks.

I would say this -

Aren't Intel X25-M's, the current darling of SSD perf, also MLC, and also subject to this issue?

Unless you have actual benchmarks showing significant slowing of a specific example drive over time, then you shouldn't make those claims. Has any respected site, like tomshardware, shown this to be the case?

 

[tubbymac]

Aren't Intel X25-M's, the current darling of SSD perf, also MLC, and also subject to this issue?

Unless you have actual benchmarks showing significant slowing of a specific example drive over time, then you shouldn't make those claims. Has any respected site, like tomshardware, shown this to be the case?

Yes the Intel drives are affected by the same thing as it's how all MLC drives must operate. The big reputable hardware sites (like Anandtech) do know how the MLC technology actually works and do report this. The honest ones try to fill the entire drive before doing benchmarks so that all or most of the cells have been written to at least once. The less informed or dishonest sites do benchmarks on the first write cycle of the cells, greatly exaggerating the performance numbers.

I do not recommend average users filling the drive before benchmarking though. That's just using up write cycles just for the sake of getting more reasonable numbers that you can already find on the honest hardware sites. No point reducing the lifetime of your SSD drive.

 

[dgdosen]

Also benchmarking on an SSD is misleading when you first get the drive. All MLC drives allow you to do a single write operation to a SSD memory cell the first time it is used. When you first get a drive obviously all the cells are unused and you will get super fast benchmarks. On EVERY subsequent write to a cell that has been written to before all MLC ssd drives require a read operation first, followed by either a write or an erase + write operation. This will significantly reduce performance and benchmarks.

Sure, I don't think any of the reviews are hiding the issues there were with single controller SSDs, nor the natural native speed of MLC vs. SLC. The issue is how the drives perform, and how they rate over time. I could care a less if G.Skill "hacked" the titan with two sub-par controllers, as long as they work well (they sure seem to) and as long as they last (that might be a question mark)

So, do us a favor, and post those bad benchmarks to this forum. I looked in other forums and other reviews, and can't find them. I think we need to see benchmarks over someone stating that performance drops in a comment...

 

[UltraNEO*]

Thanks for all the info!! It's been a really interesting read.

For me, I'll wait till there's decently sizes in a SLC SSD before I'll consider investing, I think reliability gonna be more important than speed! BTW, It's gotta be a minimum of 500GB!

 

[dgdosen]

Newegg price INCREASE?

I paid 494 for mine - and now Newegg is showing 549 with a current $20 discount to 529...

I thought prices of these things were supposed to drop like a rock - not rise. Is the Titan the new SSD X25?

 

[UltraNEO*]

I paid 494 for mine - and now Newegg is showing 549 with a current $20 discount to 529...

I thought prices of these things were supposed to drop like a rock - not rise. Is the Titan the new SSD X25?

X25 Mainstream maybe, cause it's based on MLC Technology and from the price point, it's definitely no Extreme.

 

[toolbox]

Great job, nice review nice pic's Great stuff. Always interesting to see the macs opened up

 

[tubbymac]

Sure, I don't think any of the reviews are hiding the issues there were with single controller SSDs, nor the natural native speed of MLC vs. SLC. The issue is how the drives perform, and how they rate over time. I could care a less if G.Skill "hacked" the titan with two sub-par controllers, as long as they work well (they sure seem to) and as long as they last (that might be a question mark)

So, do us a favor, and post those bad benchmarks to this forum. I looked in other forums and other reviews, and can't find them. I think we need to see benchmarks over someone stating that performance drops in a comment...

I'll do better than that. I'll actually explain how it works so you and everybody who reads this can have a deeper understanding of the main issue, which as you stated above is "how the drives perform, and how they rate over time."

First I don't want to make it sound like I'm trying to rain on anybody's parade here. Great work to the original poster, spaceball, in providing his impressions, reviews, pictures, and benchmarks. Now for the hopefully layman's explanation of the issue.

The first time you write an ssd cell it takes approximately 900 microseconds since only a single write operation is involved. Any subsequent write to that cell requires a read, an erase, and a write (sometimes the read can be skipped depending on how intelligent the ssd controller is at caching, but for simplification I'll ignore that). A read takes 50 microseconds, an erase takes 2000 microseconds, and a write again takes 900 microseconds as before. Add those up and you get 2950 microseconds. So the performance ratio is 2950 divided by 900 = about 3.28. In other words, subsequent writes will be about 3 times as slow as the very first write on an ssd cell.

Note that subsequent reads will be just as fast as the first time though as no matter when you read a cell, you only perform the read operation (no erase or write required).

So roughly expect performance to stay the same for read speed, but to degrade by about 3x when all the cells of an ssd have been written at least once.

See this link here to verify the numbers I used and to understand in greater detail how an SSD works:

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=2

See this link for details of the JMicron stuttering issue when it was first discovered:

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=7

Finally see this thread here for evidence of the stuttering and for verification of my explanation on ssd performance degradation as it pertains to the titan drive in particular:

http://forums.anandtech.com/message...ORDFRM=&STARTPAGE=1&FTVAR_FORUMVIEWTMP=Linear

 

[spaceballl]

tubbymac,

Your point is indeed well taken in regard to how subsequent writes will be slower than the initial write to the SSD. But in real-world use cases, obviously read speed is much faster. The read speed is what gives a computer "the snappy," not the write speed. Write speed is impressive for theoretical benchmarks and surely there are cases where having the fastest write speeds will be highly beneficial, but read speed is the key metric.

In regard to SSD lifespan, you do make a good point there. I guess I just don't really care that much. Computers don't last forever and in my case, I usually sell them on eBay every couple years to raise money to put toward the next one. I'm not a corporation looking to house my data for 12 years on this thing.

Lastly, as for this "stuttering" issue, I have seen no evidence of it on my system. Reading that Anandtech article, it looks like he just found lag when doing certain tasks like typing and opening safari tabs. If there is one thing I have not yet experienced since moving to this new SSD, it's lag. I'll keep my eye open for it though. The thread you linked to on anandtech was a bazillion pages long and I don't care to read through the whole thing. The stuttering issue in the anandtech article, though, was not regarding this drive, it was regarding the previous gen OCZ drive. So even though they both have a jmicron controller in it, they are utilizing them differently as the Titan does some type of RAID 0 inside. I don't think it can be said that those results are 100% relevant to the Titan.

 

[CTechKid]

When you make an argument like the one above, you need to give a baseline so the layman can draw his own conclusions. At the end of the day you've outlined an inefficiency with the technology.... but that inefficiency will still amount to the majority of MLC SSDs currently on the market out performing most if not all mechanical drives by a considerable margin. So while I appreciate the lesson in memory dynamics, I'm not quite sure why you look so unfavorably upon the MLC technology given the mechanical drive alternative.

The early adopters in the SSD market aren't ever going to hit the wear and tear wall with their SSD because they'll continue to evolve with the trend and upgrade long before the "5 year" threshold hits. I can understand your dislike for the JMicron controller, and preference for SLC over MLC, however, you're not the only with these preferences. If SLC was affordable, MLC would be an afterthought, sadly not all computer hobbyists are wealthy, so we make do with the gadget and upgrades we can afford.