IDF: SSD Myths and Hardcore Gamers

  • Thread starter Thread starter Guest
  • Start date Start date
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
so then are they going to try to get a drive around 50-75 gigs for OS and games?

or say a 36.7(?) gig hard drive for games.

Thats what i'm looking for. Just a nice simple fast drive for OS and games, and i'll use a 5400 1tb drive for movies/music etc.
 
While the chart showed the Intel drives using more power, it also shows that it has a much higher IOPS score, and thus finishes the task in less time So in the end, the Intel drive has a much better power per IOP scores, proving to be quite an efficient drive. Again, while the numbers are impressive, and we hope the new drives live up to the hype, look for us to throw up some numbers against named SSD competitors as soon as these new drives are released.
Click to expand...

Why would we trust you after the article you published? It's pretty obvious to all of us that SSDs are more energy efficient, and the article you wrote went against that, even though you had been saying it for almost two years. I'm more likely to believe Intel at this point.
 
JonnyDough sez

>>It's pretty obvious to all of us that SSDs are more energy efficient...

I know everybody really, really WANTS to believe in a dramatic power advantage for Flash SSD, but the truth is that 2.5" HDD uses only 0.6w to keep the platters spinning (0.05 watts on spin down), and disk activity drives the number up only to about 2.0 watts.

Intel's own spec sheet says 2.4watts.

http://download.intel.com/design/flash/nand/extreme/extreme-sata-ssd-datasheet.pdf


IDC's recent benchmarks have proven the original THD article essentially correct. The best and most efficient SSD's provide at best a marginal reduction in power use, and some (like Intel's) actually use MORE power.

And oh BTW, don't forget that if you look at watts-per-gigabyte, hard disks BLOW SSD AWAY. A 320GB 7,200 RPM 2.5" hard disk is TEN TIMES more power efficient than SSD when you compare power-per-GByte.

Meanwhile your CPU and screen together are using ~20x the power of the disk (Flash OR HDD).

Sooooo....even if SSD's evolve to the point where they only use 1W, the storage device is such a small piece of the total power budget that there simply is no way for an SSD to have much impact on battery life at all!!!

Finally, there's an interesting thing about silicon that I think all you guys are ignoring. Silicon power consumption INCREASES as the feature size decreases -- so unless the NAND flash makers can change the laws of physics, these devices will continue to use MORE power as the bit-density increases. Oops.

Sorry to rain on the parade...
 
T[citation][nom]anon_reader[/nom]JonnyDough sez>>It's pretty obvious to all of us that SSDs are more energy efficient...I know everybody really, really WANTS to believe in a dramatic power advantage for Flash SSD, but the truth is that 2.5" HDD uses only 0.6w to keep the platters spinning (0.05 watts on spin down), and disk activity drives the number up only to about 2.0 watts.Intel's own spec sheet says 2.4watts.http://download.intel.com/design/f [...] asheet.pdfIDC's recent benchmarks have proven the original THD article essentially correct. The best and most efficient SSD's provide at best a marginal reduction in power use, and some (like Intel's) actually use MORE power. And oh BTW, don't forget that if you look at watts-per-gigabyte, hard disks BLOW SSD AWAY. A 320GB 7,200 RPM 2.5" hard disk is TEN TIMES more power efficient than SSD when you compare power-per-GByte.Meanwhile your CPU and screen together are using ~20x the power of the disk (Flash OR HDD).Sooooo....even if SSD's evolve to the point where they only use 1W, the storage device is such a small piece of the total power budget that there simply is no way for an SSD to have much impact on battery life at all!!!Finally, there's an interesting thing about silicon that I think all you guys are ignoring. Silicon power consumption INCREASES as the feature size decreases -- so unless the NAND flash makers can change the laws of physics, these devices will continue to use MORE power as the bit-density increases. Oops.Sorry to rain on the parade...[/citation]

Hey buddy,
The MLC drives use .06W at 0 transfer but not standby and .15W at full read/write mode.... It's the SLC drives that use 2.4W which are mostly used for Enterprise storage that requires extremely high IOPS. your argument is like saying that prop planes are much better than Jet's. you just can't win in the long run of things. remember that these SSD's are relatively new and the spinning drives have been around for ever now. Think spinning disk for archiving and SSD's for Apps and fast data.
 
@anon: Learn how to use the quote button.

TG wrote a retraction because they had their data misconstrewn. The fact was that they didn't take into account the fact that the work being done was being done in a very short amount of time. They used time as a measurement for the energy efficiency of the drives, rather than the amount of work being done.

With the way current programs are written idle time is more important than read/write for energy efficiency except in server applications where there are constant read/writes. For home use idle is far more important. If SSDs begin to be used more like RAM then idle energy use will be less important for PCs.

Of course, the OS makes a huge difference as well, as Windows has significantly more paging than Linux.

The point is, that SSDs use more energy in benchmarks because the benchmarks aren't written for them. My older gen raptor HDD's kick out more heat than my CPU and GPU combined, meaning they're highly inefficient. They are STILL a major system bottleneck. So for them to say that a cool running SSD uses more energy is just bunk. It HAS to be.

T had a great point. Today's SSDs aren't being designed for energy efficiency, they're being designed for SPEED. Hard drives are just now toting energy efficiency specs. Tomorrow, SSDs will be, and I guarantee they'll use only a portion of the energy that HDDs use. Why? Because they have no moving parts. You can shut down sectors of an SSD not in use and fire them back up in a millisecond. A hard drive has a constantly spinning platter. That was the whole idea behind hybrid drives, but hybrid drives never took off because...

because SSDs are better than hybrids HDDs and ultimately will be more energy efficient.
 
Give me a ssd with atleast 300gb under 300 dollars and I'll get one But come on one game is about 13 to 20gb and I play about 8 different games. that doesn't leave me any room for my hd movies which are about 13gb each. but make it atleast 1 dollar= 1gb and I'll buy one.
 
[citation][nom]jarnail24[/nom]Give me a ssd with atleast 300gb under 300 dollars and I'll get one But come on one game is about 13 to 20gb and I play about 8 different games. that doesn't leave me any room for my hd movies which are about 13gb each. but make it atleast 1 dollar= 1gb and I'll buy one.[/citation]
Jaenail,
The spinning drives are going to be used for Larg files for a very long time.. Think of them as what tapes used to be with respect to HD's... Now the spinning drives are the tapes that are going to hold your very larg files while the SSD hold your OS and apps.
 
[citation][nom]eklipz330[/nom]for the most part, i agree with predai wouldn't mind spending a small premium [around maybe $200] for a 40gb ssd that can make bootups and loadtimes faster[/citation]

$200? 40GB? For a proper SLC SSD? Good luck finding anything close to that before 2012.
 
I agree with MArtel80.. I wouldnt use it for "such data". And I would have a backup of th4e SSD drive on my raid 1 for redundancy! And I'd be set. But 36GB.. not gunna cut it today in desktop land
 
@Area 51: Yes, in fact prop planes ARE better than jets. No, not in terms of maximum speed or flying altitude, but in the fuel economy department, yes they are. A good turbo-prop will deliver significant fuel savings over a comparably sized jet plane. 🙂
 
good speeds. would love to have 2 of these for a raid setup. kinda wondering though. could you raid 2 of these and have a regular disk drive raided to the 2 sdd's to copy it all for reduntancy and the 2 sdd's in raid 0. not sure what raid that would be i think raid 5 , no idea. but would that be possable? 2 sdd's and one disk drive
 
For gamers, all SSD brings is shorter initial load time. It's not like it'll actually increase in-game fps. The data used is copied from drive onto ram on loading anyway. Even the fastest SSD is still far slower than ram, so not having enough ram and using page file off drive is still undesirable. Nothing's actually changed.
 
At this juncture i actually trust Intel to deliver exactly what they claim. Look at the last 2 years from Intel. Have they even once not delivered what they promised? On the CPU front they have had flawless execution and actually performed better in some cases then the even disclosed.

Now AMD on the otherhand cant even get in the same ballpark as what they "claim" tons of lies misinformation you name it. Thats untrustworthy and in my book bad businees practices. actuially borderline False advertising.

So yes the odds are that Intel will deliver these drives with those or better numbers to the market and at reasonable pricing. The old days of inflated Intel pricing is gone for now.
 
Looks like there are many ignorant people commenting this article. I'm a bit disappointed since most other places i've seen more informed comments.
Firstly, yes, intel blow many of these numbers out of proportion, and the SSDs they compare theirs to are the worst you get from the first generation. You can tell by the numbers they give you.
I myself have just bought a gaming rig, and i have 2x mtron pro 32GB SSDs in raid0 for OS, apps, and CSS+AOC. And that's all i need for it. My other games and most used files are on a velociraptor, and i have a 1TB drive for storage.
In reply to demonhorde: I'm studying at an university (just started a bachelor in chemistry, going to do a master), and i found that investing $1500 in these drives were a good idea. Just search around the web and you'll see anyone else who owns mtron SSDs agree with me. I've worked the entire summer vacation for $25/hour to pay for this rig, and i don't regret it.
I would love to know the access times of these drives, and see them benchmarked with PCmark Vantage against Mtron Pro 7500series, Memoright GT, the OCZ drives (both core v2 and ssd2), and other high-grade SSDs. As the numbers given here are a joke.
 
I'm personally looking forward to SSD's and will be watching closely how they evolve over the coming years. For now I think I'll stick to HDD's because I'm a little concerned about their reliability. Having that said though, I'm sure they'll improve massively in the future.

Does anyone remember the first LCD monitors ? They had appalling image quality, pixel response times, and were desperately overpriced. Now look at them- you'd be nuts to go back to an CRT. Same thing will happen here I think.
 
I would look at Samsung - the leader in ssd's - LONG before I would trust any spintel unqualified benchmarks run by clowns who can't answer simple questions.

spintel is trying to grab the remaining 20% of the ssd market - they are late to the game, benefit from other's mistakes, and now want to spin you a story with no declared variables - this is typical of spintel benchmarks esp. when they are losing to amd. Benchmarks can be made to say anything, spintel, and we are not buying your crawp anymore.
 
One point that troubled some in the audience was the lack of an answer to lost or potentially lost data. Because the Drive keeps track of NAND http://en.wikipedia.org/wiki/Sheffer_stroke wear and can shut down individual pages and sectors, it was confusing when the Intel staff on hand couldn’t provide confirmation on if the drive would lose the data in the dying or dead sections, or if it would be moved. Hopefully, some light can be shed on this so no one will be risking their music collection or precious family photos.
Click to expand...

Maybe you should have spent some time and actually attended some classes that addressed Technical issues. Tomshardware is becoming a site that just creates doubts for their audience due to limited research which is very irresponsible for a technical site. What’s happening to you guys? There was a separate class at IDF specifically on SSD reliability that was delivered.
The Intel SSD does include comprehensive NAND management algorithms including the use of re-mapping and replacing blocks that might fail during the life of the drive so that the user data is not at risk. SSD's are expansive but well worth it.

 
The reason why they want to get gamers hyped up on SSD drives is because they know their suckers for being top of the line in all their PC parts. The truth is you'll see minimal improvements for some time until the technology pans out. These drives aren't meant for gamers though... The true benefits will be in servers of course and then it'll migrate down to video processing and other IO intensive tasks, and then finally for start up times and fast seek times.

I'd prefer a few of these in raid though simply for the faster productivity of having my apps right there when I need them. I know it might be 10-20 seconds but it adds up and I'm one of those guys who likes to see results the second I click anything.

We will have to see how prices look in a year from now.
 
[citation][nom]JonnyDough[/nom]@anon: Learn how to use the quote button.TG wrote a retraction because they had their data misconstrewn. The fact was that they didn't take into account the fact that the work being done was being done in a very short amount of time. They used time as a measurement for the energy efficiency of the drives, rather than the amount of work being done.With the way current programs are written idle time is more important than read/write for energy efficiency except in server applications where there are constant read/writes. For home use idle is far more important. If SSDs begin to be used more like RAM then idle energy use will be less important for PCs.Of course, the OS makes a huge difference as well, as Windows has significantly more paging than Linux.The point is, that SSDs use more energy in benchmarks because the benchmarks aren't written for them. My older gen raptor HDD's kick out more heat than my CPU and GPU combined, meaning they're highly inefficient. They are STILL a major system bottleneck. So for them to say that a cool running SSD uses more energy is just bunk. It HAS to be.T had a great point. Today's SSDs aren't being designed for energy efficiency, they're being designed for SPEED. Hard drives are just now toting energy efficiency specs. Tomorrow, SSDs will be, and I guarantee they'll use only a portion of the energy that HDDs use. Why? Because they have no moving parts. You can shut down sectors of an SSD not in use and fire them back up in a millisecond. A hard drive has a constantly spinning platter. That was the whole idea behind hybrid drives, but hybrid drives never took off because...because SSDs are better than hybrids HDDs and ultimately will be more energy efficient.[/citation]

Some good points, but you know little about how laptop disks work.

Laptop HDD =

0.05w idle (same as ssd).
0.6w spinning (interface in low power mode)
1.2w - 1.8w during R/W

My point (the point I actually made) was that with a 30-40 watt overall power consumption in laptop computers, getting rid of an HDD that uses (on average, Idle and Active) about 0.8 watts is not going to meaningfully improve anyone's battery life, even if the SSD used zero power.

Do the math. 30watts for 3 hrs is 90 watt-hours (typical battery capacity). Subtract about 3hrs*1watt = 3 watt hours (typical 2.5" HDD power consumption) and all you get in a 30watt laptop is (maybe) six extra minutes -- and that would be if your SSD used NO power.

Oh, by the way,

(a) what your 3.5" WD Raptor uses is totally irrelevant, and (b) your CPU and GPU uses 10x what the Raptor uses. WD Raptor uses 10w to keep the playyers turning and 16w peak (100% R/W duty cycle) Typical CPU used 50-90 watts and a high-end GPU is a 40watt device. DRAM uses another 15-25watts. Whereever do you get your numbers?

Maybe you didn't know this but the smaller diameter platters in a laptop disk spinning at 5,400 or 7,200 RPM use exponentially less power than larger-diameter platters in a desktop-class 3.5" disk. Power consumption of rotating media increases as the 4.3rd power of diameter and the 2.6 power of RPM.

Seems everyone here STILL thinks that the HDD uses a meaningful portion of overall system power.

Ain't true...

 
Another Flash SSD myth -- faster boot times.

IDC is selling thier recent technical report and benchmark results comparing Flash SSD to spinning disk for $10,000.00 to vendors and investment analysts. Of course they are not publishing the results openly (gotta buy the report), but for those who have been watching the stock market you may have seen a number of Flash SSD stocks downgraded since IDC's report came out.

Now, although we don't have details on IDC's benchmark, journalists have characterized the results as showing a much smaller performance improvement over spinning disk than what the Flash SSD hypesters have been claiming -- and that IDC also found a number of areas where they were surprised to see HDD outperform SSD.

Well, last week IDC offered a 10% discount as a contest prize for guessing the correct answer to a question based on the benchmark results. I played the game, results below.

-----Original Message-----
From: Zina Ashkouri [mailto:ZAshkouri@idc.com]
Sent: Monday, August 25, 2008 8:28 AM
To: anon_reader@anon.com
Subject: Guess which drive boots the fastest and get a 10% discount!

Thank you for accepting IDC's challenge to guess which PC drive technology is the fastest!

a) 1st generation hybrid HDD
b) 4,200rpm traditional HDD
c) MLC SSD
d) 2nd Generation hybrid HDD
e) 7,200rpm traditional HDD
f) SLC SSD

You guessed right, the correct answer is e!

http://www.idc.com/getdoc.jsp?containerId=213285

===========================================

Now, the question you are all probably asking is WHY?

The answer is that, like MOST I/O intensive applications-in-the-REAL-world, an OS boot procedure must complete a large number of read requests that cannot even be issued by the host to the drive until a previous write command has completed and has been acknowledged back to the application (in this case, the OS). This is known as the "synchronous" I/O bottleneck. So...(a) the NAND Flash "read/write" penalty slows the whole device down whenever the workload is a mix of read and write (synchronous or not), and (b) the presence of synchronous I/O in the workload further degrades performance down to the speed of small-block writes.

And, for those who know how NAND flash works...you know that this problem will be getting worse with increasing NAND Flash bit density and not better.

That's why IDC found that system boot times were faster with 7,200RPM disk than with any of the Flash SSD's they tested.

Of course, when the SSD hypesters tout "faster boot time", they don't tell you they are comparing to a 4,200 RPM disk.

Another Flash SSD myth bites the dust!!!

Now, Intel's X25-E (the super-expensive SLC version) has a massive DRAM write-cache to try to help with slow writes. It will be interesting to see if this much-ballyhooed new Flash SSD entry with 20x more DRAM than a laptop disk will boot as fast as a 7,200 RPM disk...knowing a little about Intel I'm quite convinced that Intel would not have allowed their new SSDs to be tested.


 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom