So I just received my new Lenovo E570 laptop in the mail yesterday. It includes a Toshiba XG4 (M.2 NVMe SSD) (model number THNSF5256GPUK; see https://toshiba.semicon-storage.com/info/docget.jsp?did=55015). For daily usage, everything seems perfect. Bootup takes only a few seconds.
But running some benchmarks, I'm getting some very odd results, especially on 4K write.
I installed a secondary Kingston KC400 (SATA SSD) as personal storage and I'm running benchmarks to compare the Toshiba XG4 (M.2 NVME SSD) bootable OS drive against the Kingston KC400 (SATA SSD) personal storage.
As expected, the Toshiba outperforms the Kingston on most sequential tasks, because the M.2 interface is much faster than SATA. With random reads and writes, the two drives are more competitive, because the interface itself is no longer the bottleck. With random tasks, sometimes the Toshiba is faster, and sometimes the Kingston is faster.
But it's with 4K writes that something really bizarre is happening. With 4K write, the Toshiba is about 5 to 10 times faster than a mechanical HDD (Hitachi HDD on my old Lenovo E545), but the Kingston is about 20 to 40 times faster than the Toshiba.
I am attaching some screenshots of benchmarks. "HGST" is the Hitachi HDD in the old Lenovo E545, "THN" is the Toshiba SSD in the new Lenovo E570, and "Kingston" is the Kingston SSD in the new Lenovo E570.
I've looked on Toshiba's website to see if they have any utilities or firmware updates for the XG4. I can't find anything.
Benchmarks at http://imgur.com/a/tIgUC
Thanks
As I said, the laptop is perfectly functional for non-benchmarking daily use. Bootup (off the Toshiba) takes 5 seconds, and there's never any hanging or any hourglasses.
* Update: I discovered that if I disable write-caching on the Kingston, its 4K write benchmarks drop to the same level as the Toshiba's. But whether I enable or disable write-caching on the Toshiba makes no difference to its speed. I plan on keeping write-caching disabled for reliability purposes, so I guess in the end, the two drives will both be the same speed. It's strange that the Toshiba's speed is always as if it has write-caching disabled regardless, while the Kingston's varies, but so be it, I guess?
But running some benchmarks, I'm getting some very odd results, especially on 4K write.
I installed a secondary Kingston KC400 (SATA SSD) as personal storage and I'm running benchmarks to compare the Toshiba XG4 (M.2 NVME SSD) bootable OS drive against the Kingston KC400 (SATA SSD) personal storage.
As expected, the Toshiba outperforms the Kingston on most sequential tasks, because the M.2 interface is much faster than SATA. With random reads and writes, the two drives are more competitive, because the interface itself is no longer the bottleck. With random tasks, sometimes the Toshiba is faster, and sometimes the Kingston is faster.
But it's with 4K writes that something really bizarre is happening. With 4K write, the Toshiba is about 5 to 10 times faster than a mechanical HDD (Hitachi HDD on my old Lenovo E545), but the Kingston is about 20 to 40 times faster than the Toshiba.
I am attaching some screenshots of benchmarks. "HGST" is the Hitachi HDD in the old Lenovo E545, "THN" is the Toshiba SSD in the new Lenovo E570, and "Kingston" is the Kingston SSD in the new Lenovo E570.
I've looked on Toshiba's website to see if they have any utilities or firmware updates for the XG4. I can't find anything.
Benchmarks at http://imgur.com/a/tIgUC
Thanks
As I said, the laptop is perfectly functional for non-benchmarking daily use. Bootup (off the Toshiba) takes 5 seconds, and there's never any hanging or any hourglasses.
* Update: I discovered that if I disable write-caching on the Kingston, its 4K write benchmarks drop to the same level as the Toshiba's. But whether I enable or disable write-caching on the Toshiba makes no difference to its speed. I plan on keeping write-caching disabled for reliability purposes, so I guess in the end, the two drives will both be the same speed. It's strange that the Toshiba's speed is always as if it has write-caching disabled regardless, while the Kingston's varies, but so be it, I guess?