VirtualLarry
No Lifer
- Aug 25, 2001
- 56,519
- 10,155
- 126
Newegg review for "Q series" says they are SandForce. Which would correspond with no external DRAM cache.
Yet, (a) the TechPowerUp review shows Marvell, and (b) AS-SSD scores are far higher than what an SF-2281 should be able to come up with.Newegg review for "Q series" says they are SandForce. Which would correspond with no external DRAM cache.
Thanks, I appreciate the clarification; the reason I asked is because of this article ...
http://www.anandtech.com/show/5912/toshiba-announces-thnsnf-series-ssds-19nm-nand-is-here
I took this first article to mean that the controller is Toshiba's own, not a custom Marvell as this next article (and posts here) clarifiy.
http://www.techpowerup.com/reviews/Toshiba/THNSNH512GCST_512_GB/
This article also comments on the lack of dram cache.
Yet, (a) the TechPowerUp review shows Marvell, and (b) AS-SSD scores are far higher than what an SF-2281 should be able to come up with.
Back then all we got was a picture of an IO-Data SSD with an unknown Toshiba SATA 6Gbps controller in it, so I obviously expected it to be Toshiba's own. However, later it was discovered that it was actually an SF-2281 based controller and the shipping versions also have SandForce printed on the controller.
As a THNSNH? Those reviews are specifically for the THNSNH 128GB and 256GB models, for which at least 2 sites have what should be retail-equivalent samples showing "Marvell" under "Toshiba" (along with sequential performance that looks pretty much unheard-of for a SF-2281). Different controllers for the same model would be bad PR mojo.Toshiba sells both, SandForce and Marvell, based SSDs.
As a THNSNH? Those reviews are specifically for the THNSNH 128GB and 256GB models, for which at least 2 sites have what should be retail-equivalent samples showing "Marvell" under "Toshiba" (along with sequential performance that looks pretty much unheard-of for a SF-2281). Different controllers for the same model would be bad PR mojo.
The capacity of the SSD should be an good indicator as well.
128GB, 256GB, 512GB for Marvell.
120GB, 240GB, 480GB for Sandforce.
No, it didn't. 550MBps is about your max read speed. 2800MB/550MBps = 5 seconds. If it takes much less than that, you're not copying from the SSD. Do the same at 500MBps for writes.Copying a 2.8 GB mPeg4 Video File from RAMDisk to SSD takes approx 3 secs. Copying the same file from the SSD to RAMDisk takes approx 1.5 Secs.
Interesting - But the Transfer speeds, eg: 1.5 secs Vs 19 secs, do correlate to the Transfer Rate (Speed) relative to the Speed of the interacting Media, as I've indicated?that 1.5 seconds was to copy the <1gb of the file that wasn't already in your cache, not the whole file. The majority of it was already in your ram.
1.5s can't be transferring more than about 750MB, by your own testing. Thus, the most likely situation is what you copied the file one way, then copied it other. In that time, <=750MB of it had been evicted from cache, and actually had to touch the SSD. Or, maybe almost all of it was in cache, and it took it a second to handle all the cache-checking and copying between buffers. A large part of the source data was almost certainly already in Windows' file cache.You want me to reboot between testing media transfer rates? Can't comprehend your logic RE: Ram Sessions Vs One Time incidents.
So, is there a definitive conclusion on what controller these use? I've been considering a couple of the HDTS212XZSA 128GB drives. However, as a result of previous experiences, I intend to avoid a Sandforce based product.
By the way, anyone with this drive and has it over 1/2 full notice a slowdown on it, especially when it comes to writing to the drive?
http://www.hardocp.com/article/2013/07/03/toshiba_thnsnh_256gb_ssd_review/5#.UlHIKRCE5rs
Reviews (several) seem to show it slows down greatly on random writes once 50% or more full.
http://www.xbitlabs.com/articles/storage/display/toshiba-thnsnh_9.html#sect0
It's only temporary. It's similar to "performance mode" found in some OCZ SSDs.
http://www.anandtech.com/show/7074/strontium-hawk-240gb-review/9
Every parameter is back to normal except for the speed of random writing with a long request queue, which is only half the original level. As opposed to the Toshiba, the OCZ Vertex 4 could fully recover its speed after being filled with data by more than 50%.