|
|
| Page 2 of 2 |
DXWarlock
VIP Member
Posts: 11549
Location: Florida, USA
|
 Posted: Fri, 2nd Nov 2012 21:11 Post subject: |
|
 |
| LeoNatan wrote: | | They haven't heard of distributed computing? |
Doesn't quite work the same when its doing pages is series. or on clumps. its not much faster to be worth the expense to do a cluster of machines to process it. as the CPU load balancing and the way the ink or tiff files are made in order..(normalization, ink savings, tiff creation) and needing the pages out in numerical order, causes it to have some machines 'waiting' for work. since it cant start a new page that might take 10 seconds to do..and finished before the 5 minute front page gets done.
-We don't control what happens to us in life, but we control how we respond to what happens in life.
-Hard times create strong men, strong men create good times, good times create weak men, and weak men create hard times. -G. Michael Hopf
Disclaimer: Post made by me are of my own creation. A delusional mind relayed in text form.
|
|
| Back to top |
|
|
DXWarlock
VIP Member
Posts: 11549
Location: Florida, USA
|
| Posted: Fri, 2nd Nov 2012 21:15 Post subject: |
|
|
So they do it more in a 'load balance by device" situation..Rip on one machine, asura on one machine, eco (ink) on another..so the loads taken off the main box..but still does the steps in 'series' across multiple servers.
if that makes sense.
-We don't control what happens to us in life, but we control how we respond to what happens in life.
-Hard times create strong men, strong men create good times, good times create weak men, and weak men create hard times. -G. Michael Hopf
Disclaimer: Post made by me are of my own creation. A delusional mind relayed in text form.
|
|
| Back to top |
|
|
LeoNatan
☢ NFOHump Despot ☢
Posts: 73310
Location: Ramat HaSharon, Israel 🇮🇱
|
| Posted: Fri, 2nd Nov 2012 21:20 Post subject: |
|
|
I still don't understand why tiff creation can't be paralleled.
|
|
| Back to top |
|
|
DXWarlock
VIP Member
Posts: 11549
Location: Florida, USA
|
| Posted: Fri, 2nd Nov 2012 21:26 Post subject: |
|
|
Cost vs reward. each machine added isn't 100% return on process time. not many will spend 10 grand on a server (and 10 more grand on a backup server for failover), to get good speeds.
Then spend 10 grand on another rackmount (and 10 more for backup cluster for it)to get 50% faster than what they did before.
The way it is now, if each task and software is its own island machine (rip/eco/tiff/etc) they can at least justify the cost of a new dedicated box for each task.
They rely more on reliability and predictability of tight deadlines..hence the main and 'backup' for each box, the backup being an exact hardware match of the main redundancy out the ASS at a lot of these places.
To add in load balancing across boxes that if some fail without a full backup set of matching machines, that changes the forecasted deadline time...screws up the whole process.
They are all about working as close to the "trucks arrive at 5am for the paper, it will be off press at 4:50am to be loaded".
Sure a failing set of cluster machines is a 1 in a 10000 chance..but some papers never been late to delivery in 100 years..they don't want to take a chance of having one day that breaks that record.
-We don't control what happens to us in life, but we control how we respond to what happens in life.
-Hard times create strong men, strong men create good times, good times create weak men, and weak men create hard times. -G. Michael Hopf
Disclaimer: Post made by me are of my own creation. A delusional mind relayed in text form.
|
|
| Back to top |
|
|
LeoNatan
☢ NFOHump Despot ☢
Posts: 73310
Location: Ramat HaSharon, Israel 🇮🇱
|
| Posted: Fri, 2nd Nov 2012 21:36 Post subject: |
|
|
Parallelism is rarely true 100% distribution, but it gets close to that. Especially so, in a task such as what you mention, which is rather independent. With better software design (or better software use), there can even be a better design, where you add more machines to a pool and tasks are performed faster. This method, when you are not satisfied with the result, you add more machines to the pool and improve the overall system, whereas you'd have to replace the entire computing cluster every time you wish to upgrade.
Such pools also guarantee much larger safety net, as when one machine dies, the pool is still operational.
Sounds more like technological conservatism, perhaps?
|
|
| Back to top |
|
|
DXWarlock
VIP Member
Posts: 11549
Location: Florida, USA
|
| Posted: Fri, 2nd Nov 2012 21:41 Post subject: |
|
|
Oh im not saying its not possible. Just telling you the guidelines and reasons they all give for it. We can only make our software to sell to them, if it makes them feel 'safe' by what they want, so they buy it.
They rather spend money on a machine, then a backup of it that sits idles until needed for 'comfort reasons' to make sure they dont miss a deadline..than go the route of a cluster of machines that give a faster process, but if one or 2 dies, the whole timeline is thrown off.
And as bad as newspapers are struggling now to stay afloat..the higher ups in the company cant be convinced to dish out money for a rack of cluster machines, AND a copy rack for failover for it.
You know those people that have their main Harddrive, then a external backup mirror of it, THEN a cloud mirror of it, and THEN burn it all to DVD to back it up just incase?...Newspapers are the hardware/server version of that...lol
-We don't control what happens to us in life, but we control how we respond to what happens in life.
-Hard times create strong men, strong men create good times, good times create weak men, and weak men create hard times. -G. Michael Hopf
Disclaimer: Post made by me are of my own creation. A delusional mind relayed in text form.
|
|
| Back to top |
|
|
DXWarlock
VIP Member
Posts: 11549
Location: Florida, USA
|
| Posted: Fri, 2nd Nov 2012 21:43 Post subject: |
|
|
| LeoNatan wrote: | .
Such pools also guarantee much larger safety net, as when one machine dies, the pool is still operational.
|
Yea thats the catch to them though, one machine dying without a failover one to replace it would be domino effect. since the process time increases slightly, the deadline for editorial would be changed, then the deadline for platemakers would be changed..etc.
10 minutes difference in total time for them they FREAK out...I get calls all the time about things thats a 5 minute delay, that they are acting like the buildings on fire..lol
-We don't control what happens to us in life, but we control how we respond to what happens in life.
-Hard times create strong men, strong men create good times, good times create weak men, and weak men create hard times. -G. Michael Hopf
Disclaimer: Post made by me are of my own creation. A delusional mind relayed in text form.
|
|
| Back to top |
|
|
|
|
| Posted: Fri, 2nd Nov 2012 22:55 Post subject: |
|
|
if price is no concern then why are you getting a asrock mobo since it's a lesser brand for budget pcs?
|
|
| Back to top |
|
|
|
|
| Posted: Sat, 3rd Nov 2012 00:02 Post subject: |
|
|
ok you have referred to this as a super/exceptional/kick-butt workstation, but its a high end gaming rig. a quad core and a 690 will not be able to hold their own at work. here is a quick list of suggestions:
-get more cores
-pick a motherboard that can take 64 gb of ram
-nvidia makes workstation cards for graphics as well as math/computing that will destroy a 690 at 'work' tasks
-get a fast raid card for your storage drives if you are working with large files. consider a second one for your operating system, and/or a ram drive for your most used applications
|
|
| Back to top |
|
|
thudo
Posts: 6309
Location: Mellonville North, Canada
|
| Posted: Sat, 3rd Nov 2012 05:24 Post subject: |
|
|
| Intel_NVIDIA wrote: | | if price is no concern then why are you getting a asrock mobo since it's a lesser brand for budget pcs? |
No.. its rated one of the fastest boards right now especially for OC'ing which I plan to milk. Its not on the list for general s&gs.. its there because its be reviewed to be exceptional. Boards right now that house multiple cores was never in-scope for this endeavor.
| Quote: | -get more cores
-pick a motherboard that can take 64 gb of ram
-nvidia makes workstation cards for graphics as well as math/computing that will destroy a 690 at 'work' tasks
-get a fast raid card for your storage drives if you are working with large files. consider a second one for your operating system, and/or a ram drive for your most used applications |
I'm trying to build SMART, not stratospheric.. this would utterly bloat the budget when technology is advancing at such a rate built in obsolescence is now beyond moot. I'm trying to keep this real based on whats coming down the line in the MID-TERM.
MSI GT72S 6QF Dominator Pro S 29th Anniversary Intel i7 6820HK @ 4.0Ghz, 32GB DDR4-2133 RAM, 2x256GB Raid0 Toshiba NVMe 2.5 inch PCIe SSD, Nvidia Geforce GTX 980 OC'ed 200+ Core / 200+ Mem, 17.3 inch LG IPS HD Display @ 75Hz, Intel 7265AC Wifi, Windows 10 Pro BIOS version: .112 EC Firmware version: .105
Current Broadband speed record: 329.1 Mb/sec down // 21.73 Mb/sec up
http://www.dslreports.com/speedtest/3933292.png |
|
| Back to top |
|
|
|
|
| Posted: Sat, 3rd Nov 2012 05:35 Post subject: |
|
|
i just built a system on a rIVe, 3930k, 32gb, ssd, 6tb storage, 680 top, 1kw seasonic psu (+case, mouse/kb/fans/dvd etc) for 3500 cdn (shipped+taxes). you said you were looking to stay under 5k. take out my 680 and that leaves you tons of room for some kind of workstation card, ram and a raid card (and more storage). for 5k you could probably fit a pretty sweet monitor in there too (i already had a nice one so there is no monitor included in my 3500 total).
|
|
| Back to top |
|
|
thudo
Posts: 6309
Location: Mellonville North, Canada
|
| Posted: Sun, 4th Nov 2012 16:06 Post subject: |
|
|
Yeah I want to get a nice 30" screen with resolutions above the stupid 1920x1080 limits: thats likely an extra $1-1.5K easily but my next system comes first.
Hmm.. yeah I prefer to go 22nm rather than 32nm with your 3930K choice (I want newer architecture) but I do admit the 3930 will bench better  Futher, I want a the Samsung 340 Pro for BOOT/OS for pure start-up speed: remember, I am going from 2xWD Raptors (circa mid-'06) to an SSD which alone is like night/day.
MSI GT72S 6QF Dominator Pro S 29th Anniversary Intel i7 6820HK @ 4.0Ghz, 32GB DDR4-2133 RAM, 2x256GB Raid0 Toshiba NVMe 2.5 inch PCIe SSD, Nvidia Geforce GTX 980 OC'ed 200+ Core / 200+ Mem, 17.3 inch LG IPS HD Display @ 75Hz, Intel 7265AC Wifi, Windows 10 Pro BIOS version: .112 EC Firmware version: .105
Current Broadband speed record: 329.1 Mb/sec down // 21.73 Mb/sec up
http://www.dslreports.com/speedtest/3933292.png
|
|
| Back to top |
|
|
thudo
Posts: 6309
Location: Mellonville North, Canada
|
| Posted: Wed, 7th Nov 2012 21:46 Post subject: |
|
|
| Paintface wrote: | | Perry Rhodan wrote: | | @op: the power supply is massive overkill, enermax power supply calculator says 483w for your system, so methinks 700w is enough to be on the safe side. Unless you're planning on getting a second vidcard, of course. |
+1 get the corsair 750AX or so |
Just curious you mentioned the Corsair 750AX.. Any reason not to get the Enermax (EPM850EWT) Platimax 850W 80PLUS Platinum Certified Modular PSU?
MSI GT72S 6QF Dominator Pro S 29th Anniversary Intel i7 6820HK @ 4.0Ghz, 32GB DDR4-2133 RAM, 2x256GB Raid0 Toshiba NVMe 2.5 inch PCIe SSD, Nvidia Geforce GTX 980 OC'ed 200+ Core / 200+ Mem, 17.3 inch LG IPS HD Display @ 75Hz, Intel 7265AC Wifi, Windows 10 Pro BIOS version: .112 EC Firmware version: .105
Current Broadband speed record: 329.1 Mb/sec down // 21.73 Mb/sec up
http://www.dslreports.com/speedtest/3933292.png |
|
| Back to top |
|
|
| Page 2 of 2 |
All times are GMT + 1 Hour |
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB 2.0.8 © 2001, 2002 phpBB Group
|
|
 |
|
