Tuesday, August 27th 2013

Intel to Demo SSD Overclocking at IDF 2013

It could soon become possible to overclock the controller and NAND flash of your SSD, if Intel has its way. The company is set to demonstrate how to overclock Intel-branded SSDs using its Xtreme Tuning Utility (XTU), at IDF 2013, which goes underway this September. The item on Intel's IDF itinerary marked "AIOS001" deals with seminars on overclocking Intel's next-generation HEDT (high-end desktop) platforms. X-bit Labs believes Intel could talk about SSD overclocking during that session.

Options to tweak SSDs were discovered when poking around the code of an unreleased XTU version. XTU is a unified software utility by Intel, which lets you tweak CPU, memory, and system cooling on systems running Intel Desktop Boards. Among the things end-users should be able to tweak, apart from the controller clock-speed, are the NAND flash bus-speed. Taking away interface overheads and other round-offs, 560 MB/s appears to be the practical maximum bandwidth SATA 6 Gb/s SSDs have been able to achieve. It could always be handy getting your SSD a few dozen more MB/s sequential speeds at the expense of stability.
Sources: MyCE, X-bit Labs
Add your own comment

28 Comments on Intel to Demo SSD Overclocking at IDF 2013

#1
FR@NK
WTB: Heatsink w/fan for my SSDs

:/
Posted on Reply
#2
Prima.Vera
I wonder if liquid cooling is on the way ;)
Posted on Reply
#3
RCoon
Forum Gypsy
Overclocking something that contains data is a no-go for me. I'll stick with gpu/cpu/memory thanks...
Posted on Reply
#4
Fiery
FinalWire / AIDA64 Developer
They should much rather work on eliminating the stupid bottleneck of the SATA 6.0 Gbps interface. Overclocking SSDs will lead to similar issues than when in the old days people overclocked the FSB with VLB IDE controllers (anyone remembers those?), and the data on HDDs started to get "mysteriously" corrupted. No, thanks, not again :)
Posted on Reply
#5
MicroUnC
Sounds like another marketing move!
Posted on Reply
#6
Vinska
"Look guys! I've OC'ed my SSD and got several more megs a second of write/read speeds! ...I can't find half of my files, though. Seems to have mysteriously disappeared somewhere. I suppose I'm just a bit tired and will find everything tomorrow after a good night's sleep."


by: RCoon
Overclocking something that contains data is a no-go for me.
You mean, non-volatile data? :P
Posted on Reply
#7
iO
Sounds like the PR guys came up with some super-unique feature by taking some line of products, cripple them artifically via downclocking and only let them run at full speed when paired with their pricey high-end chipsets...
Posted on Reply
#8
Fourstaff
by: Fiery
They should much rather work on eliminating the stupid bottleneck of the SATA 6.0 Gbps interface.
Working on it: NVM Express and Sata 3.2. NVM Express uses PCI-E, while Sata 3.2 is a slight increase over Sata 3.2.

by: iO
Sounds like the PR guys came up with some super-unique feature by taking some line of products, cripple them artifically via downclocking and only let them run at full speed when paired with their pricey high-end chipsets...
You always run things at most stable clock, not fastest possible. If they want to cripple their products via downclocking they are welcome. Just bear in mind that semiconductor industry is has wafer thin margins (pun not intended), and small performance increase will completely tilt the favour to the other company.
Posted on Reply
#9
Agility
And comes the day where companies start lowering the warranty coverage and increases the cost on the "featured" overclocking power of their SSD.
Posted on Reply
#11
TheLostSwede
You guys are clearly not following what's been going on in the SSD scene :p

It's about PCI Express based SSD's. Overclock the PCI Express bus = moar SSD performance.
Posted on Reply
#12
FeƤnor
by: TheLostSwede
It's about PCI Express based SSD's. Overclock the PCI Express bus = moar SSD performance.
Since when overclocking the pci-e bus is a good thing? The last three intel cpu generation (sandy, ivy and haswell) are all limited to 5-10% max bclk variation, mostly because the pci-e bus really doesn't like to be set at anything but 100 mhz.

And i don't think the pci-e will be the bottleneck, as to this day nothing comes close to saturating a pci-e 4x slot! (and don't tell me that 20 ssd in raid 0 can, you'll just prove my point...)
Posted on Reply
#13
Svarog
If there is anything i wouldn't wanna f**k with it's this.

Data is too valuable to screw with.
Posted on Reply
#14
WithoutWeakness
Last line of the article:

"It could always be handy getting your SSD a few dozen more MB/s sequential speeds at the expense of stability."

That's an absurd statement. It's not at all "handy" to raise data throughput at the expense of stability. Everything in a PC should always be 100% stable under maximum load. The last thing you should be tweaking to the point of instability is your storage system.
Posted on Reply
#15
Svarog
They should focus on making affordable 1TB+ SSDs for Storage with just 100 MB Read/Writes, can't be that hard.
Posted on Reply
#16
NdMk2o1o
by: Svarog
They should focus on making affordable 1TB+ SSDs for Storage with just 100 MB Read/Writes, can't be that hard.
They have HDD's for that, what's the point in an SSD if it's slower than a mechanical hard drive?
Posted on Reply
#17
Aquinus
Resident Wat-man
by: NdMk2o1o
They have HDD's for that, what's the point in an SSD if it's slower than a mechanical hard drive?
Ever done 4k random reads/write on a spinny disk? They're slow. SSDs handle random reads/write vastly better than their HDD counterparts. So even if the sustained read/write speed were to be lackluster, I seriously doubt that access times would be. Also anything that has moving parts is more likely to fail then something that doesn't. SSDs tend to use less power then your normal 3.5" HDD as well.

So no, I can think of a number of reasons why I would want an SSD, for reasons other than bandwidth.

On topic: I think that the idea of overclocking non-volatile storage is a terrible and asinine idea. It's not like it becomes volatile if it can't do it, it just doesn't work. Which is terrible for something that is supposed to be reliable and persistent.
Posted on Reply
#18
theoneandonlymrk
by: Prima.Vera
I wonder if liquid cooling is on the way
aha a uncovered niche in waterblock land take my godam money already, i dont cares if im not oc'ing or oc'ing it i neeeeds more pipes :roll:

oooh, oc pciex bus = revo go quicker , how did that slip my mind, to the bench,, ahoy
Posted on Reply
#19
james888
by: theoneandonlymrk
aha a uncovered niche in waterblock land take my godam money already, i dont cares if im not oc'ing or oc'ing it i neeeeds more pipes :roll:

oooh, oc pciex bus = revo go quicker , how did that slip my mind, to the bench,, ahoy
Watercool your psu
Posted on Reply
#20
Vinska
by: james888
Watercool your psu
Watercool the cables and the case.
Posted on Reply
#21
james888
by: Vinska
Watercool the cables and the case.
You can buy a watercooled power supply. No joke. It could be argued that you can cool a case with any block though.


This one is for any 3.5in hdd.
Posted on Reply
#22
Vinska
how about the cables?
Posted on Reply
#23
xorbe
Hold my beer, watch this.

What could possibly go wrong?

Look how fast I can store useless corrupted data!

Another source for random number generation.
Posted on Reply
#25
Aquinus
Resident Wat-man
by: Hayder_Master
it's worth to overclock ?
I love the smell of corrupted data in the morning! Don't you?!
Posted on Reply
Add your own comment