SSDs - State of the Product?

67MP PSD with 17 Layers, Final Size 550MB - 1 minute 41 seconds.

It is even more frustrating on an i7 (4 cores, each with hyperthreading)...13% CPU utilization!
 
ddrueding said:
67MP PSD with 17 Layers, Final Size 550MB - 1 minute 41 seconds.

It is even more frustrating on an i7 (4 cores, each with hyperthreading)...13% CPU utilization!
Click to expand...

Druderuing, I dontn thiinkso. It is 4 CPUis with Hype threading, I bet one core is fully utilizated. No way it could be 8 times faster. Maybe 5 timers?
 
Last edited:
ddrueding said:
67MP PSD with 17 Layers, Final Size 550MB - 1 minute 41 seconds.

It is even more frustrating on an i7 (4 cores, each with hyperthreading)...13% CPU utilization!
Click to expand...

I have no way to directly compare that to anything, but Adobe should get its act togther on multi-threading.
 
Fastest 939 I found was 225 dollars, Opteron 185, and that was a while ago. Results from people making the upgrade was it ran hot, and wasn't THAT much of an upgrade. IIRC, folks suggested they would have been better off moving to a newer MOBO/CPU combo.
 
The fastest "mainstream" 939 CPU I'm aware of is the x2/4800. There's probably some Opteron chips that are faster but it was the tip-top of performance before AMD moved to AM2.

Those systems are about three years old now, and I'm seeing capacitor failures on some of the lower-cost boards I used at that time, particularly MSI. It's almost like clockwork, really.

It's actually somewhat tragic, since a dual core X2 with 2GB RAM is actually a perfect desktop machine in all other ways, no matter what CPU socket it's using.
 
All my Gigabyte 939 motherboards are showing significant capacitor issues. Of the six only two will run at all, and they take multiple trys just to power up the machines. However, if they can get up and running, they will generally stay that way for weeks. Post is a real killer here.
 
P5-133XL said:
All my Gigabyte 939 motherboards are showing significant capacitor issues.
Click to expand...

Gigabyte boards have mostly been fine. The other two brands I bought heavily in that period, Biostar and MSI, seem to have plenty of issues with caps. I'm really thinking I want to see solid caps on everything at this point. Capacitors have been an issue for years.
 
I just got done looking up the tests for the 4800, and, in real world stuff, it's about 10-20%, at least according to the reviews comparing it to the 3500...
 
LunarMist said:
Perhaps most of my software does not appreciate the quad cores enough.
Click to expand...
I don't think that my software does either, but if you run enough unappreciative software at once, you might change your mind.

Not until I got to four cores did I fully realise how incredibly slow my storage subsystem was.
 
Mercutio said:
Gigabyte boards have mostly been fine. The other two brands I bought heavily in that period, Biostar and MSI, seem to have plenty of issues with caps. I'm really thinking I want to see solid caps on everything at this point. Capacitors have been an issue for years.
Click to expand...

These were low end Gigabyte MB's. I suspect that that everything changes at the bottom. Even good comapnies, like Gigabyte have to compete and sacrafices have to be made and in this case it was long-term reliability...
 
ddrueding said:
Gigabyte is pretty straightforward about it. You want a UD3 (ultra durable). More copper in the mobo, better mofsets, solid caps, etc.
Click to expand...

I'm not disagreeing, but one does pay a premium for that. Also, when I bought the MB's, Gigabyte did not have marketting that identified reliability yet. So for the most part, other than price, one didn't really know what MB's were designed for reliability and which ones were not. I just took a gamble, by going cheap whille relying on brand name, and lost in this instance. Luckily, those particular machines purposes were replaced by a couple of intel quads, running VM's so I could easily deal with their deaths.

It really made me a big believer in the value of server consolidation via VM's. The cheap machines dying actually saved money. The monthly electricity drop in letting those machines die has already paid for the replacement computers. So all is well.
 
I've found that there aren't that many cases for expensive machines that last a long time. Most applications (home machines, office workstations, laptops) do better with a machine that costs half as much replaced twice as often. I don't like it that way (more work, more pollution), but instances where a Supermicro/Tyan/Gigabyte UD3 system are financially justified are few and far between.
 
ddrueding said:
I've found that there aren't that many cases for expensive machines that last a long time. Most applications (home machines, office workstations, laptops) do better with a machine that costs half as much replaced twice as often. I don't like it that way (more work, more pollution), but instances where a Supermicro/Tyan/Gigabyte UD3 system are financially justified are few and far between.
Click to expand...

There is a premium to UD3, but it is not that much (maybe $80-100) on the scale of the total system cost. For the most part I'm willing to pay that for a MB that will last more than two years: I really try to buy for the long haul. But yes, you are correct that low-priced and more frequent replacement is often a better long term strategy because there is too much change in the industry.

It's the server MB's whos prices are unacceptable. I find it intolerable to pay $400-$800 for the right to use an overpriced Xeon or Opteron. Sometimes one just has to pony up, even when you know that you're getting ripped off because you don't have a choice.
 
P5-133XL said:
It's the server MB's whos prices are unacceptable. I find it intolerable to pay $400-$800 for the right to use an overpriced Xeon or Opteron. Sometimes one just has to pony up, even when you know that you're getting ripped off because you don't have a choice.
Click to expand...

In that case, I would question whether two lower priced boxes running desktop hardware would be a better choice. This is the Google way, right?
 
timwhit said:
In that case, I would question whether two lower priced boxes running desktop hardware would be a better choice. This is the Google way, right?
Click to expand...

Generally yes, but sometime you just can't split the workload up and you just have to buy the bigger single box. The cost to deveop custom software required to split the workload is cost prohibative except for the likes of google.
 
ddrueding said:
I've found that there aren't that many cases for expensive machines that last a long time. Most applications (home machines, office workstations, laptops) do better with a machine that costs half as much replaced twice as often. I don't like it that way (more work, more pollution), but instances where a Supermicro/Tyan/Gigabyte UD3 system are financially justified are few and far between.
Click to expand...

What is typical for consumers - once every three years?
 
I have three types of customers.

1. People who mess their system up all the time, and replace the entire machine when they do (1-2 years).
2. People who mess their system up all the time, and replace every 2-3 times (~5 years)
3. People who don't mess up their system, have simple tasks to do, and do them the same way every time. I have a lot of these customers, but considering they only call for a new system every 8-12 years, it isn't a lot of business.

Edit: Strangely, each of these groups feels that their profile is "normal" and to be expected.
 
Eventually, I'll have to buy one of these drives so that I can talk to people about it with some experience, but, in the mean time, I can bother you people.

What happens if one uses a an SSD drive without the fancy tricks and configuration changes? You use up you empty cells quicker, reducing performance; and, burn out cells quicker, reducing life. But is that it?

In particular, I'm thinking about what happens if I swap a laptop drive to an SSD and just image the old disk over, with no changes. What's the worst possible result?
 
If you get a good SSD (Intel, OCZ Vertex, re-brands of either), then nothing is required. I haven't done any trickery with my machine. Those tricks are mainly for making sucky SSDs suck less. I've done quite a few drive images onto SSDs without issue.
 
I've restored the C: image on an X25-E 50 times or more. It works as well as when new.
 
There will be more decent brands as vendors figure it out. The issue is that only 2 companies make good drive controllers at the moment. Once More companies make the controller chips, the price on that will come down, and the other vendors will start to make use of them.
 
This is REALLY huge, because endless 'old, out of date' systems will be given a new lease on life, slowing the rest of the industry, I think.
 
ddrueding said:
I noticed OCZ has a new version of the Vertex; the Vertex EX. Looks like faster writes mainly (210MB/s), but pretty darn expensive ($1,349 for 120GB).
Click to expand...

for $500, just get a ST 512GB SSD :p

I believe you are referring to the Summit series, they are newly announced as of yesterday :p ~$380...or more, for the 120GB, haven't seen the pricing on the 250.


http://www.ocztechnology.com/products/flash_drives/ocz_summit_series_sata_ii_2_5-ssd

Hmm, it appears SF isn't the *only* site I've ever had a problem accessing with the main ISP, can't get this one much either:

http://www.itechnews.net/2009/05/20/ocz-summit-series-ssd-with-128mb-cache/

Ninja works though:

http://electricalninja.com/?p=622

Would expect the 250 to be just slightly more expensive than similar Vertex:

http://www.amazon.com/OCZ-OCZSSD2-1...f=sr_1_22?ie=UTF8&s=pc&qid=1242945333&sr=1-22

Way too many different OCZ drive names/series to keep track of :(

Toshiba is supposedly about to ship their laptops with their 512GB SSD, haven't seen any real pricing on the drive alone.

What happened to the Sandisk G3 drives?
 
Handruin said:
Hot hardware has a review of several SSD drives and their performance under windows 7.0 RC
Click to expand...

http://hothardware.com/Articles/Windows-7-Disk-Performance-Analyzed/?page=9
Sure, lots of the new user interface elements are pulled from the Mac OSX design guidebook... At this point, everything seems like it's moving in the right direction with this new operating system, and Microsoft is finally showing that it can better compete in terms of usability and user-experience in today's computing environments against OSX
Click to expand...

Heh, only took them a decade to catch up :p? Get a Mac/Snow Leopard :D

EXTREME MOBILE COMPUTING:
Dual SSD RAID 0 boot drive
in MacBook Pro 'Unibody
http://www.barefeats.com/mbpp13.html

I've always been jealous of the AlienWare Area 51 extreme laptops. For one thing, they offer dual drive RAID 0 boot option. Of course, those laptops are both heavy and thick.
Click to expand...

^presume it's for 15 & 17in MBP's...too bad Apple used up all that space in the 17in for the longer running Li-Po battery, could have had a dual drive option.

http://www.barefeats.com/quick.html

May 19th, 2009 -- OCZ Vertex SSD 120G is producing impressive numbers in our lab. We're seeing an large sustained transfers of 263MB/s READ and 210MB/s WRITE on the Unibody MacBook Pro. Small random transfers per second are 3900+ READ and 3500+ WRITE. It beats the Intel X25-E on speed, price, and capacity. Stay tuned for a full report compared to other SSDs as well as the fastest HDDs.
Click to expand...
 
It's all good. Glad he posts that stuff, so it reminds me of what a racket macs are. Charge 100% more for the video card, with a driver that doesn't work, and takes months to get right, if they ever do..

Thanks U...
 
ddrueding said:
I have three types of customers.

1. People who mess their system up all the time, and replace the entire machine when they do (1-2 years).
2. People who mess their system up all the time, and replace every 2-3 times (~5 years)
3. People who don't mess up their system, have simple tasks to do, and do them the same way every time. I have a lot of these customers, but considering they only call for a new system every 8-12 years, it isn't a lot of business.

Edit: Strangely, each of these groups feels that their profile is "normal" and to be expected.
Click to expand...

Don't know about the messing systems up part. Seems like mine seem to run, and run well, for a long time.

While I know some complain of server motherboard costs, as Splash used to point out, you usually don't have much in drivers to worry about, and, they generally just plug and play, provided you use the right components, ram, etc.

There does seem to be sort of a sweet spot on Xeons, where for around 250-300 dollars you get a processor that's competitive with the consumer processors, but, allows dual or more processors.

My experience with mine has been what, 9 years now!? and, they are still going strong.

Gaming rigs seem to be replaced more often, thanks to the new advances in Direct X, etc, and the components that are avaliable.

Not considered very often is the PCI bus speed, and the memory speed. It's usually so speed delayed by hard drives, that it hasn't been a factor. Flooding the bus doing data transfer was not something that happened often.

The biggest bottle neck in computer speed has been storage, but, that seems to be changing. With SSD's, you should now be able to raid them, and flood a slower bus.

I'm already planning a new machine. As soon as SSD's get down to a reasonable price per gigabyte ratio, I don't see anything that will keep the average consumer from Raid 0 boot setups. As I was typing this, I was wondering if SSD's end up giving the same ideal sweet spot of two drives raided that SCSI does, but, I doubt it. I noticed even through data transfer time was faster, access time was not as good when 4 or more drives are raided.
I don't think this will be the case with SSD's.

Windows 7 already is better by about 25% with SSD's then the released OS. That may change, but, it does indicate that the OS might now be a limit in how quickly data is processed.

So, with SSD's I see no reason an average consumer can't run a Raid 0 with at least 2, if not 4 or more, SSD's.

Here is one area a server motherboard is a better bet. When you buy one, you hope, and it's generally the case, that the chipsets used are the best around, and, you are going to get as fast a PCI bus as it's rated, or close, rather then some of the junk motherboard chips Apple or Dell have used, that aren't close to industry standard.

I guess my real question is how fast can SSD's go, and, at what point does the system bus become a limitation?

As for period of time between upgrades:

I've noticed my single processor gaming rig, and it's group of processors, are more likely to use more CPU then the dual Xeons doing the same task. In other words, I've only got one task that comes close to using 100% of the processors on the Xeons, and, rarely if ever find a multiple combination of tasks it can't do.

The same cannot be said of the Athlon 3200 setup.
 
You must log in or register to reply here.
Back
Top