Build a new system with an i7-8700K or Upgrade to Xeon E5-2696v4's

[spork schivago]

Hello,

I read the sticky about writing a proper post, but I have some brain damage from when I was in the Marine Corps and it greatly affects my memory. So bear with me if this isn't properly formatted.

I have an HP ProLiant DL380 Gen 9 that supports the E5-2xxx Xeon processors. I have a little over 3,000$. I need your professional opinions on something. I have purchased a copy of SolidWorks Pro 2018 and it was recommended to purchase the Boxx Apexx S3, which has the I7-8700K (around 5,500$ for the options SolidWorks suggests).

Anyway, instead of doing that, I'm either going to build my own system or upgrade my server to support SolidWorks. My question is, even though the i7-8700K

https://ark.intel.com/products/126684/Intel-Core-i7-8700K-Processor-12M-Cache-up-to-4_70-GHz

has a higher core clock speed, would the DL380 Gen 9 with one Xeon E5-2696v4, the OEM version of the E5-2699v4

https://ark.intel.com/products/91317/Intel-Xeon-Processor-E5-2699-v4-55M-Cache-2_20-GHz

be faster for SolidWorks?

Here's my thinking. The i7-8700K supports a maximum of 2 memory channels, whereas the Xeon supports 4. The Xeon has a lot of cache, the 8700k not so much. SolidWorks does most of its work single threaded, but the multiple cores will come in handy during rendering. The memory bandwidth is much higher on the Xeon.


Finally, I have an HPE Smart Array P440ar SAS Controller in my server.

https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-c04496202#N1004A

Would 12Gbps SAS drives in a RAID10 configuration be faster than an M.2 PCI-E NVMe x4 drive? With the 8700K, we have 16 PCI-E lanes, but with the Xeon, we have 40. I will be using an nVidia Quadro P2000 or an nVidia Quadro P4000, more than likely.

The 8700K has an unlocked multiplier and SolidWorks recommends running it overclocked at 4.8GHz. My thinking is if I setup 4 12Gbps SAS drives in a RAID10 configuration, have four sticks of 2400MHz ECC load reduced RAM, and the Xeon, SolidWorks would run much faster on Windows 10 64-bit. What do you guys think?

Thank you.

 

[mdd1963]

Do you already have a functional server, with a cpu already installed?
If you already own the server, I'd try it as is; if your primary important application is Solidworks, perhaps it will do fine, and there would be no need for a 2nd system that most recommend for gaming. (As I've not used nor heard of Solidworks, it'd be difficult to say which system would be better for your needs, but, give your existing system a try first, unless trying it means an overly expensive Xeon CPU to do so...

4 drives in RAID 10 is merely a pair of mirrored RAID 0's, and, despite having peak transfer rates of 12 GB/sec, a pair of spinning drives can really only sustain 300-500 MB/ sec.
(An NVME M.2 drive is good for a peak transfer of 3200 MB/sec.... The impact will depend on how dependent Solidworks' performance is to local storage speeds...)

 

[spork schivago]

Do you already have a functional server, with a cpu already installed?
If you already own the server, I'd try it as is; if your primary important application is Solidworks, perhaps it will do fine, and there would be no need for a 2nd system that most recommend for gaming. (As I've not used nor heard of Solidworks, it'd be difficult to say which system would be better for your needs, but, give your existing system a try first, unless trying it means an overly expensive Xeon CPU to do so...

I do have the server, but not with a fancy Xeon, just a cheap Xeon. A E5-2603v3. Right now, I also have a HP xw8600 workstation, with a Quadro FX4800, 32GB of RAM, and two SSD in a RAID0, connected via an LSI HBA (3000 series, with a 8061) and it's running too slow for our needs.

With the server, I'd have to purchase 1) more RAM, 2) a faster Xeon processor, 3) a Quadro P series GPU accelerator.
So, that's why I'm here. Can I get faster speeds upgrading the DL380 Gen 9, or by building a 3,000$ workstation with something like the i7-8700k?

4 drives in RAID 10 is merely a pair of mirrored RAID 0's, and, despite having peak transfer rates of 12 GB/sec, a pair of spinning drives can really only sustain 300-500 MB/ sec.
(An NVME M.2 drive is good for a peak transfer of 3200 MB/sec.... The impact will depend on how dependent Solidworks' performance is to local storage speeds...)

Yes, RAID10 is a pair of mirrored RAID 0's, and because I have a true hardware RAID, that would mean the controller can write to all four hard drives at the same time. I was under the impression that these SAS drives where sustained transfer rates, and not burst, I didn't think they had a peak transfer rate, like a normal SATA drive. These 12 Gb/s SAS drives are SSD SAS drives, so there wouldn't be any spinning up I wouldn't think.

From talking to the SolidWorks tech support people, the most important things are core speed of the CPU, 32GB of the fastest RAM I can grab, and the absolute fastest hard drives I can afford. They said I could waste the money on a P4000, but in all reality, they recommend the P2000 and although the P4000 would provide a bit better performance, they don't think I'd ever notice it and it'd be a waste of cash there.

The idea of using a very expensive multi-core Xeon was because (I know this is gonna sound horrible), but I was hoping, if I did it right, I could actually run SolidWorks in ESXi / vSphere (a VM). The server currently runs CentOS 7. I was going to try to do the VM thing. I know RAID is not an excuse for backups, and I thought it'd be real nice to be able to just backup the VM container or something every once in a while.

So, to test the server, I'd have to invest a lot of money, to upgrade it. In the end, I'll do it, if it actually means it'd be faster than the i7-8700k overclocked to 4.8GHz, like Boxx does with their Apexx S3. SolidWorks Tech recommends the 8th gen 8700k, but with only 16 PCI-E lanes, the Quadro P2000 would be running at 16x, the M.2 would be running at 4x. I see a bottleneck right there. Then the 8700k's DMI3, (direct media interface (DMI), which is a point-to-point interconnection between an Intel integrated memory controller and an Intel I/O controller hub on the computer’s motherboard), versus the Xeon's 2696v4, with the QPI (Quick Path Interconnect (QPI), which is a point-to-point interconnect between the CPU and the integrated memory controller.), I thought the server with the 12Gbps SAS SSD drives and the proper Xeon CPU might actually outperform the i7-8700k with the M.2.

Just wanted confirmation before I invested what's a lot of money to us.

Thanks!!!!!