Tech EPIC WIN Thread
- Thread starter DarkAudit
- Start date
About that. I ran that benchmark with VBS enabled. After watching the latest Hardware Unboxed where they benchmarked Windows 10 vs 11 vs 11 with VBS, I learned that the average home user doesn't need VBS. And their results with the same test prove it. VBS tanks the random reads and writes performance. So here's the same test without VBS.
Glad they made the BIOS update to allow the Zen 3 Ryzen CPUs to run on B450 motherboards. The 5800X was on sale for $369 at my local Best Buy, with curbside pickup available an hour before the doors opened proper. Even though Micro Center had it for way less, the cost of gas and travel time made it a no-brainer to get it here instead. Installation was simple thanks to only having to remove the one bracket off the Core Frozr L cooler (Gamers Nexus approved, BTW). I hedged my bet by ordering a tube of Thermal Grizzly from Amazon (which hasn't arrived yet), but the Corsair TM30 I bought alongside the CPU is doing fine.
Speaking of which, the 5800X is the hottest runner of the Zen 3 lineup, due to it having all 8 cores on a single die. So I was a bit concerned seeing 89C temps on a multi-core Cinebench run. But further investigation reveals this is normal behavior, and is within spec. Plus, it was running with a fresh application of the thermal paste. It takes time and multiple thermal cycles to cure and settle in. Subsequent runs are getting incrementally cooler peak temps.
The 2600 memory controller didn't like trying to run 4 sticks of RAM at the XMP2 setting of 3200MHz. XMP1 was fine, though. The 5800X has no such trouble. 4 sticks of 8GB DDR4-3200 at XMP2 chugging happily along.
I still hope to move the new CPU to it's own build, and put the 2600 back for yet another Gentoo box, just to see how it does on actual modern equipment.
Speaking of which, the 5800X is the hottest runner of the Zen 3 lineup, due to it having all 8 cores on a single die. So I was a bit concerned seeing 89C temps on a multi-core Cinebench run. But further investigation reveals this is normal behavior, and is within spec. Plus, it was running with a fresh application of the thermal paste. It takes time and multiple thermal cycles to cure and settle in. Subsequent runs are getting incrementally cooler peak temps.
The 2600 memory controller didn't like trying to run 4 sticks of RAM at the XMP2 setting of 3200MHz. XMP1 was fine, though. The 5800X has no such trouble. 4 sticks of 8GB DDR4-3200 at XMP2 chugging happily along.
I still hope to move the new CPU to it's own build, and put the 2600 back for yet another Gentoo box, just to see how it does on actual modern equipment.
New computer is assembled, up, and running. Everything is new now except for the GPU, which is an RX580 8GB I got on sale new in 2019 for only $90 (which probably should've been an epic win in itself, looking back).
Some benchmark highlights (Old -> New):
New CPU is anywhere from 2x-5x faster, depending on workload.
280/4354 -> 617/6265 - CPU-Z CPU benchmark (single/multicore)
212/423/837/1651/2441 -> 1057/2093/3858/6523/8210 - 3DMark CPU test 1/2/4/8/16 threads
Same GPU, so similar score, no real surprise there. No longer CPU-limited, though. Probably not going to see a big change here until GPUs come down closer to MSRP and I can buy one that's based on technology newer than <checks notes> uh, 2016.
11467 -> 12593 - 3DMark Fire Strike
4242 -> 4642 - 3DMark Time Spy
Biggest improvement is SSD speed. 4x-17x faster thanks to the HUGE step up from SATA-II/300 SSD to PCIe v4 NVMe.
243/228 -> 4317/5191 - CrystalDiskMark sequential (QD=1)
24/42 -> 80/326 - CrystalDiskMark 4k random (QD=1)
The other epic win is that my estimate/budget for this build was a "modest" (by today's standards) $2500. When I added up the out-of-pocket retail cost of all the parts, the final tally came out to $2507.77.
--Patrick
Some benchmark highlights (Old -> New):
New CPU is anywhere from 2x-5x faster, depending on workload.
280/4354 -> 617/6265 - CPU-Z CPU benchmark (single/multicore)
212/423/837/1651/2441 -> 1057/2093/3858/6523/8210 - 3DMark CPU test 1/2/4/8/16 threads
Same GPU, so similar score, no real surprise there. No longer CPU-limited, though. Probably not going to see a big change here until GPUs come down closer to MSRP and I can buy one that's based on technology newer than <checks notes> uh, 2016.
11467 -> 12593 - 3DMark Fire Strike
4242 -> 4642 - 3DMark Time Spy
Biggest improvement is SSD speed. 4x-17x faster thanks to the HUGE step up from SATA-II/300 SSD to PCIe v4 NVMe.
243/228 -> 4317/5191 - CrystalDiskMark sequential (QD=1)
24/42 -> 80/326 - CrystalDiskMark 4k random (QD=1)
The other epic win is that my estimate/budget for this build was a "modest" (by today's standards) $2500. When I added up the out-of-pocket retail cost of all the parts, the final tally came out to $2507.77.
--Patrick
Last edited:
Not exactly EPIC (except that I'm actually able to use my computer again - YAY), but turns out the dreaded glitch I experienced yesterday was because the D: drive (which used to be my original HD on this 'puter) kicked the bucket. However, the fix turned out to be to just shut 'er down, unplug old drive and install a new drive - this one 1.5 TB instead of the old 1 TB Seagate.
Drives are cheap. When I first upgraded my NAS to more drive space, that left me with extra hard drives lying around. 4x4tb went into the NAS, and the 2tb drives went into my local machine. The NAS is protected by RAID. The 2tb drives act as secondary backup. I feel like it would be very unlikely for everything to take a shit at the same time and for me to lose all of my media. Whenever I'm ripping new media, I have Handbrake do the conversions and then run a .Net program I wrote that creates all of the directory structure necessary (in the case of TV shows) and then tosses the new media files to both the Plex server/NAS and to the secondary backups, so it isn't even any extra work to maintain.
