More details on DDRdrive X1: $1495 for 4Gb Version

jmke · 6th May 2009, 16:13

The DDRdrive X1 performance is not very impressive in comparison to recent SSDs because it is limited by the PCIe X1 interface. You therefore get theoretical read speeds of 250 MB/s and 155 MB/s of write. However its in terms of IOPS that the DDRdrive X1 stands out. It manages not less than 300,000 IOPS in random read and 200,000 IOPS in random write (blocks of 512 bytes). A good SSD generally gives a read of 100,000 IOPS. Moreover this level of performance is achieved at just 9.91 watts. The DDRdrive X1 can work in RAID 0.1 and 5.

This product is innovative abut a little bit limited in terms of possible usage as it costs $1495 and has a 4 GB capacity. Its availabilty has not yet been announced by we know already that the DDRdrive X1 will be guaranteed for 5 years.

http://www.behardware.com/news/10206...1-ssd-ddr.html

jmke · 7th May 2009, 01:28

looking at the Write IOPS the DDRdrive X1 is amazingly fast, faster than a system DDR drive? how is that possible

--

I-RAM... nice

4 Raptors in RAID 0...

8 Raptors in RAID 0

X25-M SSD

14gb RAMDISK

7th May 2009, 11:44

Quote:

Originally Posted by jmke

looking at the Write IOPS the DDRdrive X1 is amazingly fast, faster than a system DDR drive? how is that possible

The secret sauce to any RAM drive... build a drive controller that can at least pretend to keep up with the RAM.

Looks like they attempted exactly that but focusing purely on IOPs...

jmke · 7th May 2009, 11:53

the last result posted above is from a 14Gb ram drive (DDR3) ; it's only 12x faster than X25-M in write IOPs, the DDRdrive X1 is twice as fast as that; how can this be?

7th May 2009, 14:42

The DDRdrive's controller is hooked up to RAM, so the only limitation would be the drive controller... They obviously engineered a controller designed to better handle the speed of RAM modules, at least in IOPS.

Remember how SSD's work... X25-M uses 10 controller channels. In order to beat a system RAMdisk the DDRdrive could use mutliple channels as well, one controller per channel in "RAID-like" fashion with one or two memory DIMM's per channel. Or perhaps a single very complex controller with multiple memory controllers to simulate multiple channels. Would have to be something along this fashion.

6th May 2009, 16:13	#1
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 79,022	More details on DDRdrive X1: $1495 for 4Gb Version The DDRdrive X1 performance is not very impressive in comparison to recent SSDs because it is limited by the PCIe X1 interface. You therefore get theoretical read speeds of 250 MB/s and 155 MB/s of write. However its in terms of IOPS that the DDRdrive X1 stands out. It manages not less than 300,000 IOPS in random read and 200,000 IOPS in random write (blocks of 512 bytes). A good SSD generally gives a read of 100,000 IOPS. Moreover this level of performance is achieved at just 9.91 watts. The DDRdrive X1 can work in RAID 0.1 and 5. This product is innovative abut a little bit limited in terms of possible usage as it costs $1495 and has a 4 GB capacity. Its availabilty has not yet been announced by we know already that the DDRdrive X1 will be guaranteed for 5 years. http://www.behardware.com/news/10206...1-ssd-ddr.html __________________

7th May 2009, 01:28	#2
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 79,022	looking at the Write IOPS the DDRdrive X1 is amazingly fast, faster than a system DDR drive? how is that possible -- I-RAM... nice 4 Raptors in RAID 0... 8 Raptors in RAID 0 X25-M SSD 14gb RAMDISK __________________

7th May 2009, 11:53	#4
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 79,022	the last result posted above is from a 14Gb ram drive (DDR3) ; it's only 12x faster than X25-M in write IOPs, the DDRdrive X1 is twice as fast as that; how can this be? __________________

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Win7 already claims 5.71% of Windows Version Installed Worldwide	jmke	WebNews	3	22nd January 2010 10:02
More ATI Radeon HD 4550 512MB Details Revealed	jmke	WebNews	0	10th September 2008 12:05
Best Western Loses Details On 8 Million Customers	jmke	WebNews	0	24th August 2008 17:01
ATI Catalyst 8.8 Officially Released	jmke	WebNews	3	21st August 2008 10:04
SPARKLE Presents GeForce 7600 GS AGP PASSIVE Version to Mainstream Market	jmke	WebNews	0	21st November 2006 11:50
Pirate Windows? Buy The Retail Version at a $100 Discount	jmke	WebNews	0	15th May 2006 09:48
AMD K8 3000+ E6-Venice 1.35V Version	windwithme	Hardware Overclocking and Case Modding	1	25th June 2005 03:53
Windows XP Tweaking Companion Version 1.05	jmke	WebNews	0	11th April 2005 16:11
Windows XP Tweaking Companion - Version 1.02 Update	jmke	WebNews	0	10th January 2005 10:42
PCI Gets Version Upgrade: 12-Year Old Standard Advances Further	jmke	WebNews	0	26th April 2004 16:21

7th May 2009, 14:42	#5
Kougar Posts: n/a	The DDRdrive's controller is hooked up to RAM, so the only limitation would be the drive controller... They obviously engineered a controller designed to better handle the speed of RAM modules, at least in IOPS. Remember how SSD's work... X25-M uses 10 controller channels. In order to beat a system RAMdisk the DDRdrive could use mutliple channels as well, one controller per channel in "RAID-like" fashion with one or two memory DIMM's per channel. Or perhaps a single very complex controller with multiple memory controllers to simulate multiple channels. Would have to be something along this fashion.