Haswell build

[Shan]

The Dverb test was never meant to be "real world". It's a benchmark stress test used to measure and compare systems with and nothing more. It's like using a tape measure. It's primary use has been to see if a new build was put together and optimized properly. If we're all building the same computer and getting the same results, but a new system scores significantly lower, then we know something is way off with that build which we can now investigate. It's also used to measure different releases and versions of Pro Tools. If there's a substantial hit in a new release, then something is way off with PT and we can notify Avid. We also use it to measure and compare CPU types.

The problem with a so called "real world" test is the definition of real world itself. An individuals real world session either runs or it doesn't. Once one does a Dverb(or DaveC) test, they now have a base with which to compare their own real world sessions with. The best real world tests that we have already are the gradient DaveC tests. No sense reinventing the wheel unless someone feels the need to go from Imperial(Dverb & DaveC) to Metric.

After I do my own builds I run a Dverb benchmark to make sure everything is running smooth and it's all in the ballpark. I then come back to the forum and brag how my system is more powerful than yours and half the price. After that, I just get on with recording my real world stuff.

Shane

[YYR123]

Benchmark the ten dverb's - b/c the prior benchmark seems a bit too low for future computers

[dave911]

Originally Posted by Doobielespaul

Hey Dave. Do you think getting a d3h x87 mobo would affect overall performance or compatibility for pt10? I would shave off a bit of $ from my build.
Thanks for posting

Sent from my SGH-I337M using Tapatalk now Free

I wouldn't expect much difference unless you wanted to push the overclocking.
The current gen. of Haswell is heat limited on overclocking anyway.
I bought the 5H to get certain features.

Gigabyte has a Thunderbolt version of the 5H out now, depending on your interface might be worth the extra bucks.

[dave911]

I'll post my DVERB scores as soon as the liquid nitrogen tanks get here
Scores like this only make sense comparing the exact same system, running the exact same software, in the exact same configuration with the exact same interface.

Even for optimizing your own system you want a session that is real world for the way you work.

No you can't get any bragging rights, but you can get back to work.

[mykhal c]

Originally Posted by dave911

i7-4770K
Cooler Master V8
Gigabyte GA-Z87X-UD5H Z87 LGA 1150 2-Way SLI Dual LAN ATX Motherboard

Kingston HyperX Beast 16 GB Kit (2x8 GB) 2400MHz DDR3 PC3-19200 Non-ECC CL11 DIMM XMP Desktop Memory KHX24C11T3K2/16X

Rosewill SilentNight Series 500W 80 Plus Platinum Certified Modular Design ATX12V/EPS12V Power Supply SilentNight-500

Runs tracking at a 32 buffer with a LS56 and large (50+) plugin/track counts at 512-1024.
Overclocked to 4 Ghz static.
PT 10.3.7 an 11.02 running with no crashes or lockups.
The regular PT OS settings.

DPC Latency 30-60 microseconds.

Only issue with the motherboard is the supplied driver for the Marvell onboard disk controller is junk and throws huge DPC spikes.
The Intel controller handles 6 drives so I just disabled the Marvell, but a new driver might be available.

The memory is rated at 2400, but I and others found it runs better at XMP profile 2 -2133

If you use a FW interface this MB has a built in VIA chipset that passed the Focusrite tester.
Also has a PCI slot.

Very easy build and stable.

Some info. on why I picked this over Ivy Bridge-E
http://www.anandtech.com/show/6355/i...architecture/8

Would have waited for the 8 core Haswell -E's next year but my old system had other ideas.

YMMV!

i will add your build to the i7 sticky. i'm sure it'll catch some warranted attention!!! and i'll post the link too…good read!!!

[dave911]

OK, thanks.

I wonder how the Haswell Mac's are playing?

[The Weed]

Here's my Haswell build (also in the i7 Thread):

Case: Silverstone Fortress FT02B-USB3.0

Power Supply: Corsair RM Series RM850 ATX12 v2.31

Motherboard: ASUS Z87 PRO Socket 1150

CPU: i7 4770K Quad Core Socket 1150

RAM: G.SKILL Trident X Series 16GB (2 x 8GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) CL: 7 Model F3-1600C7D-16GTX

CPU Cooler: Noctua NH NH-D14

Video Card: Asus GT640-DCSL-2GD3 GT 640 2 GB GDDR3 SDRAM - PCI-Express 3.0 x16 Fanless Video Card Model: GT640-DCSL-2GD3

Firewire Card: SIIG FireWire 2-Port PCIe (NN-E20012-S2)

OS and Audio Drives: 2 x 500 gb WD Caviar Black SATA 6 gb/s 7200 RPM 64 mb cache drives

Backup Drive: 1 x 1 tb WD Caviar Black SATA 6 g/s 7200 RPM 64 mb cache

Firewire Card: SIIG FireWire 2-Port PCIe (NN-E20012-S2)

Software: Pro Tools 11.1.1 on Windows 8.1 Pro

Interface: 002 Rack

All Avid Windows 8 optimizations done (except Network Card Enabled) and BIOS optimizations done too

DPC Latency Checker: 7 to 13 ups for the most part with a regular spike to 997. DPCLC states the computer should be able to handle real-time audio and video streams

LatencyMon: Svchost.exe causes hard page faults, but LM states the computer should be able to handle real-time audio and video streams

Core Parking OFF: All CPUs report 100% in Task Manager

DVerb2 Score: 65 Tracks/325 DVerbs with Both Dynamic Plug-in Processing and Delay Compensation On and my Network Card Enabled (didn’t make any difference if it was Enabled or Disabled); Hyperthreading is On

System Monitor: Memory at 12%; Disk Usage 23 to 30%; Total 90 to 99%; Cores 1 through 8 running 85% to over 90%


Third Party Plug-ins: Waves Diamond Bundle v9r14, Blue Cat Free Bundle, Melodyne Single Track - just realized the DVerb2 test is supposed to be done with the Stock plug-ins only. I'll try to get some time to do the test with the third party plug-ins in the Unused Plug-ins Folder


Backup Software: Acronis True Image Home 2014

Monitor: Dell 29” Ultrasharp


Can always drop in the 8 core Haswell later.

Cheers,

[dave911]

Nice Build.
I measured my DPC with PT running and my boot up DPC is similar to yours, but I have no spikes.
My ASUS board had a Marvell disk controller that threw big spikes, you might try disabling yours if it has one.

[The Weed]

Dave,

Thanks. The spike is still in the green - just. I'm thinking it may be my KVM switch polling the USB connection. Whatever it may be, it hasn't been a problem with any sessions, including those with a video track. If it does become a problem, I'll go looking for the Marvell controller.

Cheers,

[YYR123]

A bit of fun

http://nofilmschool.com/2014/01/hack...od-tonymacx86/