ASRock Z690 Taichi review
Radeon RX 6500 XT review
MSI Vigor GK71 Sonic keyboard review
MSI GeForce RTX 3080 Suprim X 12GB review
TeamGroup A440 PRO Special 5eries 4TB NVMe review
Core i5 12400 processor review
MSI MAG B660M Mortar WIFI review
Biostar Z690 Valkyrie review
TeamGroup A440 PRO PCIe 4.0 M.2 NVMe review
DDR5 scaling with G.Skill TridentZ5 6000 CL36 review
Intel Shows Optane SSD writing at 2GB per second
Over at IDF in China a prototype of an Optane SSD from Intel has been showing some staggering numbers. The SSD is based on 3D XPoint storage technology and while copying a 25GB file it did so at 2 GB/s, less than 15 seconds.
While current nVME technology already can achieve the similar results it remains impressive none the less as this is early technology and thus just a starting point for what's to come. Intel also talks about 15TB SSDs thanks to stacked technology.
3D XPoint (3D Crosspoint) technology was announced back in 2015 with the story it could be up-to a 1000x faster than conventional NAND memory.
It looks like that the first SSDs based on this tech will be released this year already. Intel clearly mentioned that the SSD shown was an early engineering sample prototype unit. For the test setup Intel used two similar PCs with with different SSDs. The left one used traditional NAND flash memory (doh!) and the right one the Optane SSD. Both units have been connected with Thunderbolt 3.
During the transfer the NAND SSD showed a write perf of roughly 280 MB/sec (indicative for Toggle TLC NAND), the transfer for the Optane SSD to the right was roughly 1,95GB/s in a very constant flow. You can watch the IDF presentation here in this stream. The storage segment starts at roughly 49 minutes, the demo starts at 54 minutes.
Intel shows Core i7-6950X specs at 3,5GHz and 25MB cache - 04/04/2016 10:34 AM
Intel posted info on their Core i7-6950X over the weekend on their website. The 10-core processor will get a 3,5GHz base clock and 25MB cache. Other then some snippets of info, nothing specific is m...
Intel shows Broadwell, Quark and new phone SoCs - 09/11/2013 08:56 AM
At the Intel developper forum CEO Brian Krzanich exposed a preview upcoming chips. 14nm Broadwell processor will start ship to Intel's customomers at the end of the year and will be available to con...
Intel Shows New Technologies for Efficient Cloud Datacenters - 09/05/2013 08:56 AM
Intel Corporation today introduced a portfolio of datacenter products and technologies for cloud service providers looking to drive greater efficiency and flexibility into their infrastructure to supp...
Intel shows first fully patterned 450mm wafer - 01/22/2013 10:04 AM
Intel presented world's first fully patterned 450mm wafer at last week's SEMI Industry Strategy Symposium. No further details were shared by Intel at the event but previous reports indicate the chip...
Intel ships 32nm Sandy Bridge CPUs without GPU - 01/31/2012 01:44 PM
Reported a couple of times altready, Intel now has ofiically added three new IGP-less Core i5 processors to its official price list. The chips come with their GPUs disabled and can be identified by t...
David3k
Senior Member
Posts: 113
Joined: 2003-07-29
Senior Member
Posts: 113
Joined: 2003-07-29
#5259128 Posted on: 04/16/2016 12:59 PM
No the graph is not in error.
They however did deliberately cherry pick some sort of value SSD solution for comparison, likely TLC / Toggle Nand based. Writes start out amazingly fast and then drop. If you have read recent our value SSD reviews here at Guru3D.com you would have noticed very similar behavior. Once caches can't keep up, that's the performance deficit the TLC SSD will run into (for writes).
http://www.guru3d.com/articles_pages/ocz_trion_150_240_gb_ssd_review,10.html
http://www.guru3d.com/articles_pages/plextor_m7v_512gb_ssd_and_m_2_ssd_review,11.html
Also keep in mind that this "XPoint" memory is just Phase-Change Memory (PCM), and the "crossbar" design they are touting only provides marginal benefits to the overall design and performance of PCM SSDs.
http://www.theregister.co.uk/2016/04/04/memory_and_storage_boundary_changes/
http://www.theregister.co.uk/2016/01/29/xpoint_examination/
No the graph is not in error.
They however did deliberately cherry pick some sort of value SSD solution for comparison, likely TLC / Toggle Nand based. Writes start out amazingly fast and then drop. If you have read recent our value SSD reviews here at Guru3D.com you would have noticed very similar behavior. Once caches can't keep up, that's the performance deficit the TLC SSD will run into (for writes).
http://www.guru3d.com/articles_pages/ocz_trion_150_240_gb_ssd_review,10.html
http://www.guru3d.com/articles_pages/plextor_m7v_512gb_ssd_and_m_2_ssd_review,11.html
Also keep in mind that this "XPoint" memory is just Phase-Change Memory (PCM), and the "crossbar" design they are touting only provides marginal benefits to the overall design and performance of PCM SSDs.
http://www.theregister.co.uk/2016/04/04/memory_and_storage_boundary_changes/
http://www.theregister.co.uk/2016/01/29/xpoint_examination/
slyphnier
Senior Member
Posts: 813
Joined: 2009-11-30
Senior Member
Posts: 813
Joined: 2009-11-30
#5259146 Posted on: 04/16/2016 02:20 PM
i am curious about their claim 1000x more durable than NAND
after last google report about SSD
my concern about SSD durability/reliability no longer based how it will last rewrite ...
but the uncorrected errors - nand chip itself
as based google report, 20~63% drives experience at least on uncorrectable error in first 4 years
30~80% drives develop one bad block, and 2~7% develop at least one bad chip during first 4years
indeed for OS-cache usage is okay... but until they can solve how to make it more durable
ssd cant be safely used for long-time data archival purpose
now if intel means it can be rewrited 1000x more than current NAND,
then it still means nothing to me
i am curious about their claim 1000x more durable than NAND
after last google report about SSD
my concern about SSD durability/reliability no longer based how it will last rewrite ...
but the uncorrected errors - nand chip itself
as based google report, 20~63% drives experience at least on uncorrectable error in first 4 years
30~80% drives develop one bad block, and 2~7% develop at least one bad chip during first 4years
indeed for OS-cache usage is okay... but until they can solve how to make it more durable
ssd cant be safely used for long-time data archival purpose
now if intel means it can be rewrited 1000x more than current NAND,
then it still means nothing to me
Black_ice_Spain
Senior Member
Posts: 4583
Joined: 2008-05-08
Senior Member
Posts: 4583
Joined: 2008-05-08
#5259148 Posted on: 04/16/2016 02:24 PM
Damn. So fast!
But, unless you're some kind of person that requires insane drive speeds (can't think of anything off the top of my head at the moment), is speeds like this really necessary for gamers?
My 3-4 year old Samsung 840 Pro is doing me perfectly. It loads most games in 10~ seconds. Windows loads in 3 seconds. Programs and software run near instantaneous without any delay or hitching.
No, but its needed for bigdata and other things
Damn. So fast!
But, unless you're some kind of person that requires insane drive speeds (can't think of anything off the top of my head at the moment), is speeds like this really necessary for gamers?
My 3-4 year old Samsung 840 Pro is doing me perfectly. It loads most games in 10~ seconds. Windows loads in 3 seconds. Programs and software run near instantaneous without any delay or hitching.
No, but its needed for bigdata and other things
Click here to post a comment for this news story on the message forum.
Don Vito Corleone
Posts: 42895
Joined: 2000-02-22
The image shown for comparison is bull****. So why in the first picture show me a graph with 284 mb/s, while in the first few second the ssd was driving with what ???? 3500 mb / s. Don't tell that already had 284 mb/s, but the graph is in error.
No the graph is not in error.
They however did deliberately cherry pick some sort of value SSD solution for comparison, likely TLC / Toggle Nand based. Writes start out amazingly fast and then drop. If you have read recent our value SSD reviews here at Guru3D.com you would have noticed very similar behavior. Once caches and the small SLC bugger can't keep up, that's the performance deficit the TLC SSD will run into (for writes).
http://www.guru3d.com/articles_pages/ocz_trion_150_240_gb_ssd_review,10.html
http://www.guru3d.com/articles_pages/plextor_m7v_512gb_ssd_and_m_2_ssd_review,11.html