I7-990x vs 970/960 for 1080p video/CS5/BD authoring

[moviemarketing]

I'm trying to decide which CPU to buy for a new system I'm building.

It will be used primarily for the following tasks
- 1080p video editing (Avid MC5, Premiere Pro CS5) with dual monitors
- BD authoring (NetBlender, Encore CS5)
- working with extremely large outdoor advertising image files in various CS5 apps
- occasional gaming (but I want it to be able to run demanding upcoming 2011 games at 1080p, perhaps dual monitors and 3D)
- possible future upgrade for 3D BD authoring

I considered a Xeon board with multiple CPUs, but decided against for the following reason:

I know that a machine with many cores would help with video encoding tasks or perhaps authoring/rendering tasks. However, the primary editing app that I use is Avid Media Composer 5. It is a 32-bit application that has to be run in a sort of Windows XP mode. At the office I use a 12-core MacPro, and I'm not sure how much Avid or FCP are benefiting from all those cores, as FCP is also a 32-bit application, which I understand are not very optimized for multi-threading.

I hope that a 6-core i7-990x is sufficient CPU for demanding multi-threading tasks that I might need to perform on my home computer.

The difference in price between the i7-990x and the 980x is only $50, certainly seems worth it.

However, the i7-970 is $400 cheaper and the i7-960 is $600 cheaper.

What would be the best way to estimate how much time I would save with CPU intensive tasks like encoding, authoring, rendering if I spend a little more on the 990x vs. going with the 960 or 970?

 

[moviemarketing]

Thanks for your response.

I have a bit of understanding about the math behind how many cores we are talking about. However, is there a way to estimate on a practical level how much time it will save with the specific apps I'm using? Maybe some blog has done a really good benchmark of these CPUs?

The cost of Xeon boards and processors is not too bad, but none of the Xeon boards had USB 3.0 integrated and I couldn't find any with two x16 PCI-e 2.0 slots. And if I am doing encoding in Avid MC5, which is a 32-bit application, it would seem not to benefit so much from dual processors.

 

[ElMoIsEviL]

Thanks for your response.

I have a bit of understanding about the math behind how many cores we are talking about. However, is there a way to estimate on a practical level how much time it will save with the specific apps I'm using? Maybe some blog has done a really good benchmark of these CPUs?

The cost of Xeon boards and processors is not too bad, but none of the Xeon boards had USB 3.0 integrated and I couldn't find any with two x16 PCI-e 2.0 slots. And if I am doing encoding in Avid MC5, which is a 32-bit application, it would seem not to benefit so much from dual processors.

http://www.evga.com/products/moreInfo.asp?pn=270-WS-W555-A2&family=Motherboard%20Family&series=Intel%205520%20Series%20Family&sw=5

Performance
Based on Intel 5520/ICH10R chipset
Supports Dual QPI Socket 1366 Intel Xeon 5600 and 5500 Processors

Memory
12 x 240-pin DIMM sockets
Triple Channel DDR3
Maximum of 48GB of DDR3 1333MHz+
Expansion Slot
0
Storage I/O
1 x UltraDMA133
6 x Serial ATA 300MB/sec with support for RAID 0, 1, 5, 10 and JBOD
2 x Serial ATA 600MB/sec with support for RAID 0 and RAID 1
Integrated Peripherals
8 Channel High Definition
2 x 10/100/1000
Multi I/O
2 x USB3.0 ports
10 x USB2.0 ports
Audio connector (Line-in, Line-out, MIC)
FireWire 0
Form Factor
HPTX Form Factor
Length: 13.6in - 345.4mm
Width: 15in - 381mm

And I don't think PCI-Express slots are a problem on this board

 

[moviemarketing]

http://www.evga.com/products/moreInfo.asp?pn=270-WS-W555-A2&family=Motherboard%20Family&series=Intel%205520%20Series%20Family&sw=5

holy crap! thanks, you just made my decision more difficult now ;-)

Seriously that board looks perfect! I suppose I need to decide which Xeon CPUs to plug in.

 

[moviemarketing]

OP here is the Adobe CS 5 benchmarks. You can check the times and the setups used combined with disks, amount ram etc etc

http://ppbm5.com/Benchmark5.html

The E5620 doesnt come near the 980x in CS5. The 5680 beats it. The 980x is the 2nd fastest right on the heels of the 5680 Xeon pairing.

In the CPU test MPEG2-DVD the 980x beats the E5260 by 25 seconds!
CPU test H.264 BR the 980x beats it by over 30 seconds!

This is excellent and exactly the info that I needed, thanks very much for sharing this benchmark.

If I understand correctly from reading the analysis, my takeaway would be

1. dual E5680 / i7-980x would be pretty comparable

2. dual E5620 not worth the extra cost

3. i7-990x - probably best option or at least on a par with dual E5680s

If I am willing to buy 48GB RAM there would be an incremental benefit to going with a Xeon board, but it still doesn't seem to be massively outperforming the over-clocked i7-980x machines with 24GB, at least while performing these benchmark tasks.

 

[moviemarketing]

This is all excellent info!

This is a home computer and I was planning to go with single or dual SLI GTX 470 or 580 to benefit from the Mercury Playback Engine, while still being able to run the occasional game at 1080p or higher resolution, possibly 3D.

I was planning to buy a top-tier OCZ PCI-e SSD and either use external RAID for additional storage or if there is a PCI-e slot available install an Areca RAID controller for the additional storage.

I'm considering the MSI Big Bang Xpower and the Asus Rampage III .

If I understand correctly, I should be able to configure the following setups on either mobo, depending on whether I decide to go for dual video cards in SLI

PCI-e SSD @ x16
GTX 580 @ x16

or

PCI-e SSD @ x16
GTX 580 @ x8
GTX 580 @ x8

 

[moviemarketing]

Adobe make use of the disk caches. So a Sata raid config will perform better than the ssds.

The faster the disk(s), the better. Raids do improve performance. Notice that all Top 20 Performers use Raid configurations and sometimes even multiple Raids. Even SSD's, though widely touted for their speed, benefit significantly from Raid configurations.

I'm not that familiar with the PCI-e SSD tech, but my understanding is that the card I'm considering comes with an onboard RAID controller and essentially it is 1TB of flash memory in Raid0, with read/write speeds on a par with a 5x or 10x Intel SSD RAID.

I have 2 x 160GB Intel X18 G2 SSDs in Raid0 in my laptop with sequential read/write speeds around 500-550MB/s, which is pretty fast, however I believe the 1TB OCZ Z-Drive R2 P88 sequential read/write speeds approach 1.4GB/s, with 512MB cache.

Does this mean that Adobe would be able to make use of the disk cache and the performance would be on a par with an SSD Raid, or am I misunderstanding how Adobe accesses RAID storage?

If so, it would seem to be a cheaper solution (~$3700 on sale at newegg) than buying 10 x 80GB or 8 x 160GB Intel X25 SSDs + Areca RAID controller, for example.


http://redirectingat.com/?id=1402X558040&xs=1&url=http%3A%2F%2Fwww.newegg.com%2FProduct%2FProduct.aspx%3FItem%3DN82E16820227517&sref=http%3A%2F%2Fwww.tomshardware.co.uk%2Fforum%2Fmessage.php%3Fconfig%3Dtomshardwareuk.inc%26cat%3D10%26post%3D299691%26numrep%3D2253540%26ref%3D11%26page%3D1%26p%3D1%26subcat%3D174%26sondage%3D0%26owntopic%3D0%26new%3D0%23formulaire