New Build Fun...

[Maibock]

Hello, my name is Sean and I'm a geek I guess, I love the process of selecting components and putting them together almost as much as using the final product. I'm afraid I've been out of touch with the latest tech news for a while and would love some advice on a build I got going.

These are the parts I have ordered and newegg kindly delivered a day or so ago:

MS Windows Vista 64-bit Ultimate OEM
Intel Core 2 Quad Q6660 (Yay it's a slacr!!), with AS5 of course
CoolerMaster Hyper TX2 CPU Cooler (prolly shoulda went tuniq tower, but I heard good things about it)
Gigabyte EX-38-DQ6
Sapphire HD 3870X2
(2) G.Skill F2-8500CL5D 2GBx2 DDR2 (I love filling all the dimms up)
(2) Seagate 500GB Barracudas (gonna use Raid 0)
Lite-On Blu-Ray drive
Gonna use an Asus SATA DVD burner I already have

The advice I'm fishing for is basically the psu and maybe case. I already have an Antec P180 case with a Corsair 620HX PSU. I definitely plan to overclock and I could be wrong but I don't think the 620HX will be up to it. I was also thinking of going with a Stacker 830 or 832 case with a few good 120mm fans in it.

I've looked at the PC Power & Cooling 750 Quad thinking it might be an ok choice if the Corsair wasn't up to the task. Any suggestions are appreciated.

[SuperSix]

IMO - the Corsair should be fine.

Why RAID 0? Minimal gains, double the chance of losing all of your data.

And unless you plan on Crossfire in the future, that board is WAY overkill.

[madmat]

Just because it supports X-Fire doesn't make it overkill. It's actually a board that makes a lot of sense. Supports DDR2 rather than the costly DDR3, supports 1600FSB so when they roll out there's an easy upgrade path. As it supports native 1600 FSB, it'll be a good candidate for overclocking a 1333 or 1066 FSB CPU. Not to mention all it's other features.

It's not fair to judge a mobo on just it's GPU slots alone.

[Maibock]

I was thinking of crossfire in the not too distant future, I realize it's probably more mobo than I need atm, of course I definitely don't NEED 8GB of RAM either, but with ddr2 prices I was like why not. Another consideration is if I replaced the corsair psu and my antec case then I could give my current system to my parents who don't have a computer but want one just to surf the web, which it would do just fine. Just for kicks here's the current system in question:

AMD Opteron 144
DFI Lanparty Ultra (non sli version)
2GB G.Skill something or other ending in HZ
Geforce 7800 GT
2 80GB Hitachi SATA hard disks in Raid 0

[treshix]

Look, I really hate to bust in to something like this on my first post, but then sometimes I see something I have to post on, and with so much technical info floating around here, I can't help but add.

You're using Hitachi's now. Hitachi's are good, solid drives. I've got a pair that I've rolled over the SMART hours-on counter over at least once - so they're sitting at 170 thousand hours of operation now.

They have no bad sectors, don't have ECC errors, and work flawlessly to this day.

I have two "highly-rated" seagate 320's. They're a month old.

Their cables are tightly connected. They're getting appropriate power from the rails. My PSU is prolly running at about 50% load, if that, and it's a PSU that I've never had a problem with.

Multimeter tests out fine.

So, I've owned the Seagates about 3 months. There have been 172,748,596 hardware corrected ECC read errors. This means that the head does not stay aligned with the data track properly.

We have a seek error rate of 8,971,895.

The unit has 448 power on hours. It has been turned on 25 times in it's life.

The second seagate has had 150,331,316,278 hardware ECC read errors, and 165,072,671 seek errors. It has also only been turned on 25 times in its life.

The units have a number of very bad quirks, such as recording max temperature reached as 21'C, but current temperature is 31'C.

These are the same size, but are different firmware version, different lots, different PCB.

They are both very badly designed drives. They both run hot under constant forced air cooling... I'm sorry, but I have to recommend against seagates.

If I had the money to spend again, I'd be buying more Hitachi's in a heartbeat. forget this good review stuff, I'll stick with a manufacturer whose parts are proven good in server applications for my own use until such time as they are not, or I have money to burn on hdd's that might be bad. These seagates are slow, don't meet spec in a number of ways... Argh!


So, yeah. I'm sure other people have had different experiences with seagates, but how many reviewers actually poll the drive itself and ask it what's going on under the hood?


And before you ask, yes, both of these drives are Barracuda's.


Anyway, on to your question.

TDP for a slacr is 95W. Now, if it was an AMD, that'd be theoretical max power consumption, but it's not. Intel TDP has different criteria from AMD TDP.


http://www.jonnyguru.com/forums/show...p?t=885&page=6 is a very good thread to read on this particular topic (It ought to be stickied, I refer to it so often I have it bookmarked now.)

Using this thread as a baseline for "Where to find the info", we have http://download.intel.com/design/pro...s/31559205.pdf page 18ish.

We find max amps are 115 for the slacr. Now, we need to know what the voltage derivation is, at 115a - we find it at .150. What's the maximum voltage for this chip? 1.5.

So, for the slacr, (1.500 - 0.150) * 115 = 155.250W.

155.250 watts is, essentially, the theoretical maximum draw this chip can take without going poof. It's a good baseline in general to use when calculating what PSU you must have, as you're unlikely to ever get to the "Components pushed so hard they go poof" stage, and that means you're calculating in headroom.

An old engineer I knew once said "Always calculate in an extra 50%, in case you're wrong somewhere else."

And this is why you can't compare Intels TDP with AMD's TDP. AMD's TDP /is/ the theoretical max draw of the chip. Intels is a more "Average" number.

Next, since we're being picky, how much juice does your HSF want?
Cooler-Master's site actually has that info already calculated for us - 12v rail, .18A, 2.16W.

EX-38-DQ6 - Gigabyte doesn't list any power draw requirements. Thus, I'd have to look up tech specs for the main component chips, and I really don't want to do that.

I've always just said "35W max" and called it good, so - 35W max, figure 3A max draw.

3870 is rated for a maximum wattage of 95. I'm going to go out on a limb and say a 3870x2 is twice that, so a total of 190W. Total _available_ is 225 (75w from the PCIe bus, 75W from each 6pin PCIe connector), so it has /some/ room for overclocking. However, the 8pin PCIe connectors are rated for 150w. I don't have any knowledge of how greatly OC'ing would affect this.

Anyway, 190W at 16A.

4 DDR2 DIMMS, spec calls for 4.4W. F2-8500CL5D calls for 2.0 to 2.1V, so we can figure that at 2A per unit. (Somebody check my math on this, because we've also got mosfets involved here at an assumed 80% effeciency rating, but that amperage looks high to me.)

...but we'll say 8A for 4 DIMMS of it, a total draw of 17.6W (per JEDEC for DDR2 anyway), so 17.6W / 8A for RAM.

I'm assuming the SV35.2 model barracuda's for simplicities sake. OEM has the info on the site in the tech specs, the highest wattage they list is 9.80, at 2.8 amps. (I'm not convinced they have the wattages entered in the right boxes, but so long as the two highest numbers are right...) The interesting thing is the 2.8 amps. That involves a max start wattage of 33.6, assuming the unit pulls only off the 12v rail. So, 33.6 watts, 2.80A.

Lite-On's make me cringe. Why not a Plextor?

I'm going to call ballpark figure at 30W, lite-on's site (like mode DVD/rw/Cd/blu-ray/whatever sites for shiney, reflective media) provides no info on max requirements.

And another 30W for the SATA burner.

.18A 2.16W Heatsink
115A 155.250W But it's on the other side of mosfets, so call it
115A 194.0625 CPU.
3A 35W Mobo estimate.
16A 190W 3870x2
8A 17.6W Four DDR2 DIMMs.
2.8A 33.6W Single seagate 500GB drive.
2.8A 33.6W Oh yeah, you want two.
2.5A 30W Blu-Ray guesstimate.
2.5A 30W SATA burner estimate.

152.6A total, 556W PSU absolute minimum.

Of course, this assumes you shove as much power into the CPU while you're starting the computer and running 3D applications that slam the 3870x2 into walls while loading your ram 100% while burning a blu-ray and a DVD and doing video processing on the hard drive and lighting a !$%!ing cigarette.

i.e. slightly overkill. If you ever manage to even hit that, you won't stay there long - it'd be a fluke in the first place, IMHO.


Now, I personally happen to like Athena PSU's, but nobody ever reviews them, nobody else ever uses them, but I've multimetered them under load real system load and found them to be rock solid... Usually less then 2% out of spec. But that's my experiance. Everyone here A) knows what they're doing, and B) Has more experiance then I do, and C) Likes corsair/zippy pretty constantly.

Ergo, Corsair 750W. It's peak efficiency is about 375W, which is right where I'm going to peg your probable system wattage use. That gives you some extra headroom, can handle your emergency load situation if the universe should suddenly collapse, and assuming 24/7 operation under load, gives you capacitor aging headroom for about two or three years before you start to tax the PSU.

In real world experiance, I keep my PC's well cooled and it's not uncommon for me to get between three and five years operational life out of a given PSU. Usually I swap them out after about five years just on the principle.

Also, UPS and clean power will increase the PSU's lifespan on principle. I usually go high quality surge protector, and use the UPS as a signal cleaner / backup power, unless I'm dealing with a line dirty enough the high freq noise filters through the PSU badly. (Causes battery problems for laptops...)

Um.

If I've said anything really stupid, please correct me. Also, I havn't addressed ripple, which is one of my pet peeves against PSU's in general. Athena has, from my testing, an ungodly small ripple.

Anyway, lots of this is from my own personal research and experiance. Just, finally finding a site where other people actually had the same kinds of opinions I did? I had to join.

[madmat]

<snip>

You say that that Deathstar has 170 THOUSAND hours on it? So it's 19 years old?

170,000/24 = 7083.3333333333333333333333333333 days. 7083.3333333333333333333333333333/365 = 19.406392694063926940639269406393 years. There are only 8760 hours in a year.

You asked for corrections on your data.

[Jon Gerow]

I have two Hitachi's: a 1TB and a 750GB. Love them both. Incredibly cheap for such large capacity drives.

But I also have two Barracuda 7200.10 250GB's.

No problems. No errors. Nada.

And when I was building VM servers (about 10 to 15 a month for over a year) I used Seagate 80GB SATA drives exclusively. Two per machine, RAID1 software. Failure rate: about < 1.5%, although my "data" is only two years old (left the company after two years).

Maybe whoever you get your Seagate drives from doesn't handle them properly. I would get my drives typically from tier one disty in the big foam block boxes, but when I had to get one or two from a retailer like Newegg or Tiger, I would often get a drive in a plastic shell just thrown in a box of foam peanuts. I would never even bother to implement the drive. I would immediately get an RMA and complain about poor packaging.

[treshix]

Yeep...

Those numbers I pulled when I was tired. You can see it in the math. Erf.

The unit currently shows 36516 hours of usage. That's about four years. I've a note from about 3 years ago when I noticed the unit had rolled over. I assumed that it rolled at 65535, but I never did the math on it.

However, I had notated roll at 100,000 hours. I was troubleshooting an reiserFS corruption issue at the time that later turned out to be the fault of a combination of issues.

Units in question were purchased the same day in late '01 or early '02. Hitachi claims the SN's are invalid. IBM links to Hitachi to look.

Technical information on the SMART side of things is relatively impossible to find. Everyone who had one of these series exploded over having a "deathstar", but none of the GXP120's or GXP180's ever suffered from the bad run/bad plant/ possible oxidization issues that the GXP75's, which were the actual death stars, had.

I doubt most of them understood that glass platter drive technology will *shatter* if you drop it.

But I digress. My notations were incorrect. 65535 was obviously not the rollover point, and neither was 100000. (And my math on that 100000 was done horribly. erp.)

My guesstimation is that it rolls somewhere in the 40,000 range. I have smart logs run at 34k hours, followed by smart logs run at 14k hours, &c in the firmware SMART log.

I will say that these unit have been in basically a 24/7 RAID array or paired, since I purchased them in late 01 / early 02.


They still have yet to gain a bad sector.

Here's a dump of the SMART information from one of the drives:

  1 Raw_Read_Error_Rate     0x000b   100   100   060    Pre-fail  Always       -       0
  2 Throughput_Performance  0x0005   100   100   050    Pre-fail  Offline      -       0
  3 Spin_Up_Time            0x0007   108   108   024    Pre-fail  Always       -       263 (Average 244)
  4 Start_Stop_Count        0x0012   100   100   000    Old_age   Always       -       994
  5 Reallocated_Sector_Ct   0x0033   100   100   005    Pre-fail  Always       -       0
  7 Seek_Error_Rate         0x000b   100   100   067    Pre-fail  Always       -       0
  8 Seek_Time_Performance   0x0005   100   100   020    Pre-fail  Offline      -       0
  9 Power_On_Hours          0x0012   095   095   000    Old_age   Always       -       36517
 10 Spin_Retry_Count        0x0013   100   100   060    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   000    Old_age   Always       -       969
192 Power-Off_Retract_Count 0x0032   098   098   050    Old_age   Always       -       2509
193 Load_Cycle_Count        0x0012   098   098   050    Old_age   Always       -       2509
194 Temperature_Celsius     0x0002   196   196   000    Old_age   Always       -       28 (Lifetime Min/Max 14/54)
196 Reallocated_Event_Count 0x0032   100   100   000    Old_age   Always       -       0
197 Current_Pending_Sector  0x0022   100   100   000    Old_age   Always       -       0
198 Offline_Uncorrectable   0x0008   100   100   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x000a   200   200   000    Old_age   Always       -       0

SMART Self-test log structure revision number 1
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Extended offline    Completed without error       00%     36042         -
# 2  Extended offline    Completed without error       00%     33671         -
# 3  Extended offline    Completed without error       00%     14417         -
# 4  Short offline         Completed without error       00%     14416         -

As you can see, I've never really paid attention to these drives. They're "backed up", and are old enough that I use them as emergency fallbacks in case the seagates go bad.... But I've started to debate swapping the seagates out now and binning them.

For comparison, here's outputs from one of the seagates:

SMART Attributes Data Structure revision number: 10
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x000f   115   093   006    Pre-fail  Always       -       0
  3 Spin_Up_Time            0x0003   096   096   000    Pre-fail  Always       -       0
  4 Start_Stop_Count        0x0032   100   100   020    Old_age   Always       -       12
  5 Reallocated_Sector_Ct   0x0033   100   100   036    Pre-fail  Always       -       0
  7 Seek_Error_Rate         0x000f   054   045   030    Pre-fail  Always       -       150331351537
  9 Power_On_Hours          0x0032   100   100   000    Old_age   Always       -       445
 10 Spin_Retry_Count        0x0013   100   100   097    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   020    Old_age   Always       -       25
187 Unknown_Attribute       0x0032   100   100   000    Old_age   Always       -       0
189 Unknown_Attribute       0x003a   100   100   000    Old_age   Always       -       0
190 Unknown_Attribute       0x0022   072   048   045    Old_age   Always       -       605683740
194 Temperature_Celsius     0x0022   028   052   000    Old_age   Always       -       28 (Lifetime Min/Max 0/26)
195 Hardware_ECC_Recovered  0x001a   061   059   000    Old_age   Always       -       174058699
197 Current_Pending_Sector  0x0012   100   100   000    Old_age   Always       -       0
198 Offline_Uncorrectable   0x0010   100   100   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x003e   200   200   000    Old_age   Always       -       0
200 Multi_Zone_Error_Rate   0x0000   100   253   000    Old_age   Offline      -       0
202 TA_Increase_Count       0x0032   100   253   000    Old_age   Always       -       0

SMART Error Log Version: 1
No Errors Logged

SMART Self-test log structure revision number 1
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Extended offline    Completed without error       00%        40         -
# 2  Extended offline    Completed without error       00%        24         -

SMART Selective self-test log data structure revision number 1
 SPAN  MIN_LBA  MAX_LBA  CURRENT_TEST_STATUS
    1        0        0  Not_testing
    2        0        0  Not_testing
    3        0        0  Not_testing
    4        0        0  Not_testing
    5        0        0  Not_testing
Selective self-test flags (0x0):

After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.



As you can see, before I'd had these drives for 48 hours I was running smart tests because "Stuff was wrong, yo."

OEM's like seagate do not take kindly to being told that a /pair/ of drives have irregular data-read spindle speeds indicative of misaligned heads and/or internal formatting or track spacing issues. Note, that the lifetime high temperature is below the current temperature - this shouldn't be possible.

These drives are currently less then 50% full. I will be binning them as soon as I can stick Hitachi 1TB drives in there.

As for the counters themselves, doing math and trying to figure out what happened, I'm starting to wonder if the drives lose usage data if they're left powered down for too long. They have been bounced from state to state a couple of times and were once off for two weeks while I waited for the rest of the components to catch up.

Hrm. Tis a puzzle I do not know the answer to. I can guarantee they have run the last 5 years 24/7, and they were originally used in a server that was up 24/7, and I've used them in a number of 24/7 roles since then. I've no idea what the "correct" numbers are though.

I should probably do an hourly check on them and log the value.


-----

JonnyGURU: It's entirely possible the distributor didn't handle them properly. I snagged them for an emergency replacement of a WD that died after 8 months. (Controller fried. Controller was half-fried when I bought it. WD wouldn't RMA for internal thermistor recording constant temps of 70'c when drive was cool to touch. Why I ever buy *anything* but Hitachi's, I have no idea.)

Ironically, at this point I have a good enough working relationship with the company I use as a distributor, that I can walk in with any piece of equipment that has any price tag, and simply say "It's bad." and have no questions asked. Still, it is definitely not a top tier distributor.

In fact, being as I assemble my servers out of what is essentially home PC components and get roughly 99% availability out of them, I tend to be /really/ picky.

Sadly, I qualify as merely a home hobbiest. I've never had the opportunity to try and stick a server I built in a truly high load situation. Someday, maybe.

Otoh, my statistics graphs show with out a doubt that my home server is roughly 5x as reliable as my broadband ISP's ciscos are. Isn't that irony?

[madmat]

treshix, what are the particulars of the system that you're running those drives in? I ask because I had a system kill two WD 120GB drives by running the SATA ports out of spec. It was an older ABIT 865 chipset powered mobo. They had set up locks on the AGP/PCI/SATA ports for overclocking but I later found out that two of the four SATA ports weren't indeed locked and were prone to running out of spec when overclocking. For over a year or so I suspected WD drives, wouldn't recommend them nor would I buy any. Come to find out, WD were blameless in my troubles.

[treshix]

It's a GA-M55plus-S3G, not overclocked, running stock everything.

I switched to nForce mobo's after Via turned into crap.

ATI/AMD mobo's might be a better bet from my perspective soon, but I'm waiting for the SBx00 series to mature further and for the B3 stepping phenom to come out before I go that direction.

My laptop has an SB600 in it. I suspect I wouldn't have a problem if I wasn't forced to do "strange" things to make the wireless work in linux, but as it is I end up with IRQ vector response invalid / IRQ invalid interrupt vector errors and eventually, if I use 3d and wifi at the same time, end up in an IRQ cascade that is inevitably on the same CPU as my keyboard IRQ is attached to.

Classic case of "You can look but you cant touch!"

Other then that hiccup, I find the toshiba a perfectly stable little laptop. A215-S4807, if anyone cares.


For some reason, I remember part numbers of things I fight with better then the ones that have just worked like magic for years.

... Mobo before the current GB was an MSI Neo-V mobo. Northbridge was constantly generating MCE's. BIOS chip slowly went bad. Pity, it was my last AGP mobo, and it was the computer with my fancy ATI card. I bought said ATI card when that mobo was the best I had. *shrug* C'est la guerre.

Heh.