I'm upgrading my current workstation (circa 2006-2010) with a Ryzen build using the 1800x.
Current Workstation:
CPU: Intel Quad-Core 2.66Ghz 2.13Ghz
CPU Cooler: Tuniq
MB: Gigabyte
Memory: Gskill 8GB
GPU: Nvidia Quadro FX3700
OS Drive: Crucial SSD 128GB
Other hard drives: Two WD 150GB Raptors in Raid0 using 3Ware controller
Case: Lian-Li
Two Monitors: NEC LCD2070NX, Dell 3007WFP
Power: Corsair HX-620W
New Workstation:
CPU: AMD Ryzen 7 1800x 3.6GHz
CPU Cooler: NZXT Kraken x62 (Push/Pull)
MB: AsRock Taichi x370
Memory: 32GB Corsair Dominator Platinum DDR4 @2666Mhz
GPU: Nvidia Quadro FX3700 or Nvidia GTX 1070FE (or both) (will be buying the Quadro P5000 at a later date, maybe)
OS Drive: Samsung 960 Pro NVMe M.2 SSD 512GB
Other hard drives: Two WD Black 4TB, Crucial SSD 128GB
Case: Be Quiet! Dark Base 900 (not the pro)
Four monitors: NEC LCD2070NX, Dell 3007WFP (and two others, haven't decided)
Power: Corsair RM750i
Repurposing old parts:
I'll be using the Crucial SSD as a scratch disk, the Quadro GPU until I upgrade to the P5000, and the two older monitors.
Would there be any benefit in making use of the two Raptor 150GB drives with the 3Ware controller? I'm guessing modern SSDs out-perform these 10k hard drives even in hardware Raid0.
Multiple monitors with multiple GPUs:
I want to continue to use the older Quadro FX3700 card (PCI-e 2.0 x16) until I upgrade to a more recent Quadro. But, I want to add two additional monitors to the workstation.
When using two GPUs in the Taichi board, as is typical, the cards will run in 3.0 x8 mode. In my case, since this won't be a SLI setup, I am trying to figure out the throughput available when using the 1070 with the FX3700.
Based on wikipedia:
The throughput bandwidth of a PCI-e 2.0 x16 equals the bandwidth of a PCI-e 3.0 card running at x8. Does this mean that my FX3700 Quadro will run at full bandwidth in the first x16 or second x16 slot of the motherboard?
I would normally just call technical support at AsRock for such a question, but they don't provide phone support.
My initial thought is yes. But, there must be catch.
Current Workstation:
CPU: Intel Quad-Core 2.66Ghz 2.13Ghz
CPU Cooler: Tuniq
MB: Gigabyte
Memory: Gskill 8GB
GPU: Nvidia Quadro FX3700
OS Drive: Crucial SSD 128GB
Other hard drives: Two WD 150GB Raptors in Raid0 using 3Ware controller
Case: Lian-Li
Two Monitors: NEC LCD2070NX, Dell 3007WFP
Power: Corsair HX-620W
New Workstation:
CPU: AMD Ryzen 7 1800x 3.6GHz
CPU Cooler: NZXT Kraken x62 (Push/Pull)
MB: AsRock Taichi x370
Memory: 32GB Corsair Dominator Platinum DDR4 @2666Mhz
GPU: Nvidia Quadro FX3700 or Nvidia GTX 1070FE (or both) (will be buying the Quadro P5000 at a later date, maybe)
OS Drive: Samsung 960 Pro NVMe M.2 SSD 512GB
Other hard drives: Two WD Black 4TB, Crucial SSD 128GB
Case: Be Quiet! Dark Base 900 (not the pro)
Four monitors: NEC LCD2070NX, Dell 3007WFP (and two others, haven't decided)
Power: Corsair RM750i
Repurposing old parts:
I'll be using the Crucial SSD as a scratch disk, the Quadro GPU until I upgrade to the P5000, and the two older monitors.
Would there be any benefit in making use of the two Raptor 150GB drives with the 3Ware controller? I'm guessing modern SSDs out-perform these 10k hard drives even in hardware Raid0.
Multiple monitors with multiple GPUs:
I want to continue to use the older Quadro FX3700 card (PCI-e 2.0 x16) until I upgrade to a more recent Quadro. But, I want to add two additional monitors to the workstation.
When using two GPUs in the Taichi board, as is typical, the cards will run in 3.0 x8 mode. In my case, since this won't be a SLI setup, I am trying to figure out the throughput available when using the 1070 with the FX3700.
Based on wikipedia:
The throughput bandwidth of a PCI-e 2.0 x16 equals the bandwidth of a PCI-e 3.0 card running at x8. Does this mean that my FX3700 Quadro will run at full bandwidth in the first x16 or second x16 slot of the motherboard?
I would normally just call technical support at AsRock for such a question, but they don't provide phone support.
My initial thought is yes. But, there must be catch.
Last edited: