This must be slow, DD: Time to upgrade?

[Santilli]

Work2: i7-4770, 32GB RAM, 840 Pro 512GB, 3x 3TB, 2x Titan, GA-G1.SNIPER 5, H100i, AX1200i

Hint, Hint, Hint.

Does this run gmail and 7 or 8 ok?

Web?

Plus you didn't say what OS?

Seriously:

If I was to pick a cpu right now it, would be this one, or the 4770K.

Let me know what's blown up, and what you might want to move.

GS

 

[ddrueding]

I can pretty much tell you right now that your board is fried. Even with a new PSU, CPU, and RAM with everything else removed it won't POST reliably. When I can get it to POST memtest will fail despite really safe timings. Let me see what I have kicking around, I have already built the replacement to Home1

 

[Santilli]

Any deals on UPS's would be great. Both the APC and .... went south, at 8 years and 10 years....

 

[Santilli]

I really like double the speed, and half the voltage of the 4770 and 4770K....;-)

DD,. you ARE the best.


gs

 

[Santilli]

PS:

Work one looks a bit anemic. Hasn't Splash required you to do photos from space labs?;-)
gs

PS what happened to splash?

 

[Santilli]

This system was protected by both a UPS and a APC voltage reg. Kind of concerned about what made it go south....

 

[Santilli]

Yes, I get DOS....
Merc used to argue for consumer stuff to be replaced by enterprise. My experience is with enterprise hardware, this system would have run, and been effective for at least 10 years, if not more, for my uses. But, the gigabyte MOBO was designed to die in 5 years.

Merc: your advice was wrong...

 

[CougTek]

Yes, I get DOS....
Merc used to argue for consumer stuff to be replaced by enterprise. My experience is with enterprise hardware, this system would have run, and been effective for at least 10 years, if not more, for my uses. But, the gigabyte MOBO was designed to die in 5 years.

Merc: your advice was wrong...

Your Gigabyte EX58-UD5 isn't enterprise-grade hardware. It is mostly high-end consumer grade. If it made it to 5 years, then it wasn't particularly bad (or good). It is the expected service lenght of consumer-grade hardware.

If you want something more robust, I suggest something like this. The CPU(s) won't reach the highest frequencies and there is no overclock possible, but it should last longer than any bling-bling consumer-grade motherboard and power supply. Of course, you'll have to dig in your wallet too. The Xeon E5-2637 v2 aren't that expensive though. 16GB DDR3 Kingston sticks are around 180$ each too, so it's not that bad.

 

[P5-133XL]

Your problem with reliability is likely not so much the manufacturer of the MB but rather the amateur anti-static procedures that were used when the machine was built. Even one totally invisible spark can massively damage the lifespan of whatever electronics were touched without anyone ever realizing that anything was damaged. Building from scratch has a lot of advantages but long-term reliability is not one of them, unless strict OCD-like anti-static procedures are used.

 

[Santilli]

I don't think DD is really in the "amateur anti-static procedures" category as a machine builder. Originally it was DD's machine...
I do realize that was a dig at me, and thanks for that, but, you missed.

More likely, I have a motherboard that simply went south.

Out of two athlon- gigabyte motherboards, one went south in 5 years, the other I'm typing on. These built in 2006, IIRC.

The cpu's also went south on both machines. This one is a dual core, which doubled the speed of the original, for around 40 dollars, IIRC.

Trade off is Merc was right, that in one sense, a lot of hardware does change in that time.

On the other, the main functions I do don't really stress the Beast, and I hardly ever play games. I can see how many XP based business machines are still in service.

With gaming stuff, it makes particular sense to not build for long life. I can't complain. I had a very fast machine, for a long time, at a great price.

I just realized I don't really have much time to be playing games theses days...

 

[P5-133XL]

I assure you that my post was not a dig on you are DD but rather all self-made machines. Very few people or small companies, even those well informed, build machines do so in a highly anti-static environment with proper anti-static equipment properly used. Large PC/Server manufacturer's do. There is a significant difference, in reliability between buying CougTec's Supermicro workstation pre-assembled and buying the exact same components and having someone build it for you (which can be done, typically cheaper) .

I'm not arguing that server components don't have an inherent reliability difference (they do for they are tested more) but rather how everything is treated after manufacturing that matters far more.

 

[Santilli]

Do Dell, HP, and the other consumer computer makers follow such anti-static environment assembly techniques?

Any statistics to back this up, or tests, etc.?

 

[CougTek]

Do Dell, HP, and the other consumer computer makers follow such anti-static environment assembly techniques?

They probably do for their consumer-grade hardware and they most certainly do for their server-grade hardware.

Any statistics to back this up, or tests, etc.?

It's basic computer technical knowledge combined with years of experience in the field. People carry on average around 15KV of static electricity. While the current amount is insignificant for us, it is far from insignificant when it travels through copper traces and transistor of nanometers width. It shouldn't be hard too understand that even a tiny electric charge can fry a significant amount of nanometer circuitry (what's in basicly every processor, memory chip and most cihps on a modern motherboard). No need to use Google for statistics and data regarding this. As I wrote, it is basic computer knowledge.

When I worked on an assembly line for Nortel Networks, being grounded at all time when we were on the production lines zones was a requirement. You could be suppended if you were caught without anti-static straps. We used the same standard procedures as in other reputable plants where surface mount hardware was made.

 

[P5-133XL]

Big voltages just destroy which is relatively easily caught. Smaller voltages, drill small holes that do not totally penetrate the circuits and thus do not cause immediate failure but rather failure over time and that is why static matters a lot for long-term reliability.

 

[Santilli]

Thanks. Great information. I'm probably responsible for the damage to my own computers....

 

[Santilli]

Reason I haven't posted here much. You all are in a different league from my skills with computers.

Irony is some folks think I actually know something about computers, and won't pay me for other services, unless I come over and try to solve their computer problems.

 

[Stereodude]

ESD is like the boogeyman that gets blamed for the otherwise unexplainable.

 

[LunarMist]

Yeah. Who know why the board failed. It was not exactly the fastest CPU even back then, so it's time to move on.

 

[P5-133XL]

My comments were not purposed to assign blame or cause of the failure.

Santilli, wants long-term reliability and I was just trying to help him get there.

 

[P5-133XL]

Thanks. Great information. I'm probably responsible for the damage to my own computers....

Perhaps, but only if you are touching/handling circuit boards or chips. Properly using an inexpensive wrist strap can go a long ways towards preventing self-inflicted wounds. Once a machine is boxed-up, touching the case doesn't matter for it is all grounded.

 

[Santilli]

My comments were not purposed to assign blame or cause of the failure.

Santilli, wants long-term reliability and I was just trying to help him get there.

Thx.

With the way stuff is packaged, and my consumer/whatever consumption of computers, and parts packaged it seems to prevent ESD, i.e. plastic cases on memory etc., I've never thought ESD was really much to worry about, and thought the computer companies who design components had designed protection into their packaging. I can't really remember the last time I touched anything on that mobo, except plastic...

Guess I was wrong on this one....

 

[Santilli]

My comments were not purposed to assign blame or cause of the failure.

Santilli, wants long-term reliability and I was just trying to help him get there.

Thx.

With the way stuff is packaged, and my consumer/whatever consumption of computers, and parts packaged it seems to prevent ESD, i.e. plastic cases on memory etc., I've never thought ESD was really much to worry about, and thought the computer companies who design components had designed protection into their packaging. I can't really remember the last time I touched anything on that mobo, except plastic...

Guess I was wrong on this one....

 

[P5-133XL]

The packaging works, but the time handling between removal from the packaging and installation is critical. You should always have, at the minimum, a wrist strap when doing things like swapping out a CPU, adding a stick of RAM, or even upgrading a video card. Also, we are talking long-term reliability so it can be years before any damage done shows up as a failure. General rule when dealing with chips and boards: The less anything is handled, the more reliable it will be.

 

[sedrosken]

You gotta be kidding! An i7, slow?

I'll take it off your hands if you want, Sant...

I'm not picky.

Also, I agree, if I were to buy an i7 (which defies all logic I possess, being on a budget and all) I would go for the 4770k. It appears to be massively overclockable, I read about certain small pc builders overclocking them to 4.2 GHz on air as mere routine. Coupled with a TITAN or two and a decent SSD with as much RAM as you want (since money seems to be no object here) you'd have a very solid system on your hands.

 

[sedrosken]

Couldn't edit fast enough D:

I never thought ESD to be much of an issue either. Probably why my freebie RAM for my old desktop fried before I put it in. But, coming out of a scavenged computer I figured I couldn't have expected much. I mean, it was trashed for a reason right?

The toolkit that I got for xmas this past year came with T8 and T10 bits (I could finally jack into old broken HDDs *squee*) and an ESD strap (that I only used once, and it constantly got in my way). Also, a ratcheting screwdriver (though I don't often use the ratcheting part... it tends not to work quite right).

 

[Mercutio]

You gotta be kidding! An i7, slow?

The oldest i7s, the 920s and 860s, are five years old now. They're emphatically NOT slow CPUs, but at the same time a newer i5 is something like 50% faster on a per-core basis.
I don't know what Santilli is doing with that machine that he would call it slow, but I can see a huge difference in video encoding performance between my oldest i7 and my currrent day-to-day desktop.

 

[sedrosken]

... Like I said, if you don't want it and are just going to throw it away, give me a holler...

HINT HINT HINT


(Am I being subtle enough? )

 

[Mercutio]

I have an i7 920 in my office that I don't think has been powered on in two years, sed. Make me an offer.

 

[CougTek]

Also, I agree, if I were to buy an i7 (which defies all logic I possess, being on a budget and all) I would go for the 4770k. It appears to be massively overclockable, I read about certain small pc builders overclocking them to 4.2 GHz on air as mere routine.

Not massively overclockable. Hitting 4.2GHz from a 3.5GHz base clock isn't all that special. The TIM (thermal interface material) between the CPU core and the heat-spreader is shit on Haswell processors, although this will soon be rectify by Intel in the upcoming months. The TIM in mainstream Ivy Bridge processors was also weak, hence the reason they typically don't overclock as well as the older Sandy Bridge processors. I hope the 10x 2100$ Xeon E5-2690v2 we've bought 5 months ago at the office have better TIM than their consumer-grade counterparts.

 

[sdbardwick]

I hope the 10x 2100$ Xeon E5-2690v2 we've bought 5 months ago at the office have better TIM than their consumer-grade counterparts.

Yup. LGA2011 processors use solder rather than globs of TIM between the die and heat spreader.

 

[Stereodude]

The TIM (thermal interface material) between the CPU core and the heat-spreader is shit on Haswell processors, although this will soon be rectify by Intel in the upcoming months.

Where did you hear about this, and what are they planning to do?

 

[CougTek]

Where did you hear about this...

I either pulled it out of my ass, or I read it here.

...and what are they planning to do?

This new processor will be a fully unlocked Haswell core, codename ‘Devil’s Canyon’ and should be out mid-year (note, Computex is in June). The features that Intel is promoting are the improved thermal interface material, an updated packaging and 9-series chipset support.

I don't know in what consist the improvement to the TIM used in the upcoming Haswell processor.

 

[Santilli]

The 940 is plenty fast for anything I do. It also eats twice the energy of the current processors, and according to passmark, is half the speed of a 4770k. I'm not an overclocking type.

What is a bit of a bother is the new cpus, starting with the i7, five years ago, and coupled with a decent motherboard, video card, and SSD create computers that are so fast I don't really see a need to upgrade every 5 years, unless it's to compensate for Pacific Gas and Electric jacking up our electric bill, every chance they get.

Currently I am also on a budget. I NEVER would have built this machine myself. Components at the time are entering into the area of cost not justifying the performance increase.

I generally find you have a base for the common production chip, like the 920.

Intel Core i7 920 @ 2.67GHz 5007 250 17.57 $285.00
vs.
Intel Core i7 940 @ 2.93GHz 5481 219 7.83 $699.95*

So is the increase in processing power from 5007 to 5481 enough to justify the added 415 dollars in cost? Never would have bought that myself, unless I got a GREAT deal on it.

Likewise the video card, and the double drive raid 0, though being a storage nut, this is perhaps the most tempting, until you find out you can't really tell the difference between 300 mb/sec and 525 mb/sec, with near 0 ms access times.

It's going to be interesting this month to find out actually how much power I'm paying for each month with the i7 gone, and going back to the old 3800 Athlon...

 

[sedrosken]

I was more or less kidding. I can't afford that stuff. My budget is non-existent.

A donation, however...

Just kidding. I don't want to feel like I owe anyone, makes me feel weird talking to them.

 

[Santilli]

Hi Merc
Whats in the box, and what do you want for it????

email me..
Thx

gs

 

[Handruin]

I was more or less kidding. I can't afford that stuff. My budget is non-existent.

A donation, however...

Just kidding. I don't want to feel like I owe anyone, makes me feel weird talking to them.

If it wasn't a damn fortune to ship a 1U rack server from MA to MI, I'd donate one to you...but it's heavy. I think it's a dual socket-capable with one socket Xeon.

 

[ddrueding]

Bad news for Greg, this computer seems to work fine now. In the end I believe the issue was insufficient cooling. After stripping out some unused SATA and IDE cables, as well at through use of an air compressor, and replacing some fans things seem to be working fine. Two sticks of the RAM had failed, and I suspect that was the cause of the corrupted RAID0, but I have Win8.1 Pro running Passmark's super burn-in test repeatedly without issue. Granted the computer is a bit louder than it was before, but things are working fine.

 

[timwhit]

Bad news for Greg, this computer seems to work fine now. In the end I believe the issue was insufficient cooling. After stripping out some unused SATA and IDE cables, as well at through use of an air compressor, and replacing some fans things seem to be working fine. Two sticks of the RAM had failed, and I suspect that was the cause of the corrupted RAID0, but I have Win8.1 Pro running Passmark's super burn-in test repeatedly without issue. Granted the computer is a bit louder than it was before, but things are working fine.

Shouldn't you be running F@H on it for burn-in purposes?

 

[ddrueding]

The Passmark tool produces the highest system-wide power draw of anything I've used, going so far as to read from any optical drive, play audio, STR the drives, and constantly refresh the video. For systems with questionable stability, it is the best I know.

 

[Handruin]

Prime95 configured with high memory usage does a great job of weeding out issues also. It doesn't touch GPU or disk subsystem but for my needs it has been a great tool.