Craving some AMD goodies? 3950X's inside :) PLUS Holiday Giveaway!

cybertech

Crazy crazy vps with ECC ram. Saving up real hard for one!

poisson

@dahartigan said:
The first thing that strikes me, I have no idea why the CPU shows up as "AMD EPYC" when it's a Ryzen. This could be some hypervisor thing, and it really makes no difference because the clock speed matches what the Ryzen is, and the results speak for themselves.

I suspect because it is a very new Ryzen, the kernel may not be updated. Try using one of the newest kernel and see if it reports correctly.

Delong

@cybertech said:
this is madness

@masedi said:
Is there EPYC 2 vps?

Do we have a 7002 series? Not yet. Hopefully we'll have some by this time next year though!

@dahartigan said:

Nice! =D

I just unwrapped this beauty and ran a bench because I've been dying to see how high of a geekbench score I could get on a server. The things we do to amuse ourselves, eh?

This is the L version of this Ryzen 3950X VPS package from Bandit Host.

The first thing that strikes me, I have no idea why the CPU shows up as "AMD EPYC" when it's a Ryzen. This could be some hypervisor thing, and it really makes no difference because the clock speed matches what the Ryzen is, and the results speak for themselves.

Check out that freaking score. 5972 single core performance and with 4 threads you get 18462 which checks out. Holy shit that's fast. Kind of blowing my mind fast.

This is definitely a premium VPS. Excellent CPU performance, excellent network (I've had no glitches since being a customer) and the support is fantastic.

I'm still trying to get my head around how freaking fast this thing is. I think my next experiment will be to temporarily shift my Plex server there, which is currently on a 12 vCPU EPYC VPS, and see how it performs. My average utilization on the EPYC is around 15-20% so this would definitely put it through its paces (and probably push the boundary of fair use, so I'll really need to consider how long the test should run)

If there was anything to complain about it would be the confusion with the CPU showing as EPYC, but that's too trivial to even bother me, just that it's literally the only negative I could think of lol

In closing, I think this is an extremely awesome product and I'm really excited to see how the product lineup evolves during 2020 given the 2 offers so far have been amazing AMD gear with a really decent network and great interactions with support. I've kicked myself in the past for not getting in early with new providers after the prices go up, so my advice is get these while they're hot.

Thanks for the bench and kind words!
If you give your VPS a reboot it should start popping up as a 3950x now haha

dahartigan

@Delong said:

@cybertech said:
this is madness

@masedi said:
Is there EPYC 2 vps?

Do we have a 7002 series? Not yet. Hopefully we'll have some by this time next year though!

@dahartigan said:

Nice! =D

I just unwrapped this beauty and ran a bench because I've been dying to see how high of a geekbench score I could get on a server. The things we do to amuse ourselves, eh?

This is the L version of this Ryzen 3950X VPS package from Bandit Host.

The first thing that strikes me, I have no idea why the CPU shows up as "AMD EPYC" when it's a Ryzen. This could be some hypervisor thing, and it really makes no difference because the clock speed matches what the Ryzen is, and the results speak for themselves.

Check out that freaking score. 5972 single core performance and with 4 threads you get 18462 which checks out. Holy shit that's fast. Kind of blowing my mind fast.

This is definitely a premium VPS. Excellent CPU performance, excellent network (I've had no glitches since being a customer) and the support is fantastic.

I'm still trying to get my head around how freaking fast this thing is. I think my next experiment will be to temporarily shift my Plex server there, which is currently on a 12 vCPU EPYC VPS, and see how it performs. My average utilization on the EPYC is around 15-20% so this would definitely put it through its paces (and probably push the boundary of fair use, so I'll really need to consider how long the test should run)

If there was anything to complain about it would be the confusion with the CPU showing as EPYC, but that's too trivial to even bother me, just that it's literally the only negative I could think of lol

In closing, I think this is an extremely awesome product and I'm really excited to see how the product lineup evolves during 2020 given the 2 offers so far have been amazing AMD gear with a really decent network and great interactions with support. I've kicked myself in the past for not getting in early with new providers after the prices go up, so my advice is get these while they're hot.

Thanks for the bench and kind words!
If you give your VPS a reboot it should start popping up as a 3950x now haha

root@ryzentest:~# cat /proc/cpuinfo | grep "model name" | uniq
model name : AMD Ryzen 9 3950X 16-Core Processor

That's much better! Thanks for fixing that so quickly, a working example of that awesome customer service I mentioned :-)

dahartigan

I'm extremely impressed at how strong these cores are on the plow. I would say that 4 cores of this beast is the equivalent in Plex terms as 8 cores of most intel processors.

The other very impressive thing is the full 1 gbit speed from google drive is possible, I haven't used many networks where that's the case, for example LA seems really bad.

Haven't seen full usage of the Plex server yet, but over the next 24 hours I will see how it performs :-)

saibal

@dahartigan said: I haven't used many networks where that's the case, for example LA seems really bad.

You mean pulling from GDrive into a LA node?

poisson

@dahartigan said:
I'm extremely impressed at how strong these cores are on the plow. I would say that 4 cores of this beast is the equivalent in Plex terms as 8 cores of most intel processors.

The other very impressive thing is the full 1 gbit speed from google drive is possible, I haven't used many networks where that's the case, for example LA seems really bad.

Haven't seen full usage of the Plex server yet, but over the next 24 hours I will see how it performs :-)

I have very, very compelling evidence that his network is exothermic potassium to APAC. Will be done testing in Jan 2020 and you will see what I mean.

gtxr

I got Ryzen-L-VPS too
The CPU is very good.

Region: Global https://bench.monster v.1.4.9 2019-12-24
Usage : curl -LsO bench.monster/speedtest.sh; bash speedtest.sh -Global

---

OS : CentOS 7.7.1908 (64 Bit)
Virt/Kernel : KVM / 3.10.0-1062.1.1.el7.x86_64
CPU Model : AMD Ryzen 9 3950X 16-Core Processor
CPU Cores : 4 @ 3493.434 MHz x86_64 512 KB Cache
CPU Flags : AES-NI Enabled & VM-x/AMD-V Enabled
Load Average : 0.00, 0.01, 0.05
Total Space : 50G (7.7G ~17% used)
Total RAM : 3789 MB (286 MB + 3312 MB Buff in use)
Total SWAP : 4095 MB (0 MB in use)

Uptime : 0 days 14:53

ASN & ISP : AS18450, WebNX, Inc.
Organization : WebNX, Inc.
Location : Los Angeles, United States / US

Region : California

## Geekbench v4 CPU Benchmark:

Single Core : 5964 (EXCELLENT)
Multi Core : 18378

## IO Test

CPU Speed:
bzip2 : 146 MB/s
sha256 : 975 MB/s
md5sum : 572 MB/s

RAM Speed:
Avg. write : 3754.7 MB/s
Avg. read : 10752.0 MB/s

Disk Speed:
1st run : 1.1 GB/s
2nd run : 1.0 GB/s
3rd run : 1.1 GB/s

---

Average : 1092.3 MB/s

## Global Speedtest

Location Upload Download Ping

Speedtest.net 901.58 Mbit/s 935.36 Mbit/s 10.092 ms
USA, New York (AT&T) 171.07 Mbit/s 263.87 Mbit/s 79.754 ms
USA, Chicago (Windstream) 229.48 Mbit/s 373.52 Mbit/s 39.506 ms
USA, Dallas (Frontier) 195.38 Mbit/s 509.25 Mbit/s 44.413 ms
USA, Miami (Frontier) 206.52 Mbit/s 289.24 Mbit/s 73.746 ms
USA, Los Angeles (Spectrum) 374.00 Mbit/s 722.60 Mbit/s 22.802 ms
UK, London (Community Fibre) 173.63 Mbit/s 257.75 Mbit/s 119.128 ms
France, Lyon (SFR) 58.71 Mbit/s 17.17 Mbit/s 137.483 ms
Germany, Berlin (DNS:NET) 117.13 Mbit/s 188.15 Mbit/s 159.144 ms
Spain, Madrid (MasMovil) 125.51 Mbit/s 7.43 Mbit/s 140.433 ms
Italy, Rome (Unidata) 122.62 Mbit/s 222.55 Mbit/s 156.736 ms
Russia, Moscow (MTS) 57.91 Mbit/s 150.87 Mbit/s 195.841 ms
Israel, Haifa (013Netvision) 26.50 Mbit/s 167.02 Mbit/s 188.909 ms
India, New Delhi (GIGATEL) 24.93 Mbit/s 87.66 Mbit/s 274.753 ms
Singapore (FirstMedia) 51.48 Mbit/s 29.99 Mbit/s 191.865 ms
Japan, Tsukuba (SoftEther) 174.48 Mbit/s 313.60 Mbit/s 104.110 ms
Australia, Sydney (Yes Optus) 105.94 Mbit/s 75.36 Mbit/s 166.598 ms
RSA, Randburg (Cool Ideas) 11.68 Mbit/s 14.85 Mbit/s 283.529 ms

Brazil, Sao Paulo (Criare) 67.33 Mbit/s 204.87 Mbit/s 161.033 ms

Timestamp : 2019-12-27 14:52:23 GMT

• https://browser.geekbench.com/v4/cpu/15078336

dahartigan

An update for anyone interested. I have moved my Plex over to the Ryzen L to see how it performs and I have noticed that things are much faster, especially with transcoding. The CPU usage is actually significantly lower than when run on Intel processors, as is the system load.

I was shocked to see that without any change to the usage pattern, this seriously outperforms everything else this has run on. The crazy part is my 24 hour CPU usage average is 9% for 4 threads of Ryzen whereas previously it was 15% on 12 EPYC threads with the same load.

I think I'll continue this experiment for a month and update my findings in a week, perhaps I need to work on a Plex benchmark to somehow quantify this.

poisson

@dahartigan said:
An update for anyone interested. I have moved my Plex over to the Ryzen L to see how it performs and I have noticed that things are much faster, especially with transcoding. The CPU usage is actually significantly lower than when run on Intel processors, as is the system load.

I was shocked to see that without any change to the usage pattern, this seriously outperforms everything else this has run on. The crazy part is my 24 hour CPU usage average is 9% for 4 threads of Ryzen whereas previously it was 15% on 12 EPYC threads with the same load.

I think I'll continue this experiment for a month and update my findings in a week, perhaps I need to work on a Plex benchmark to somehow quantify this.

The Ryzens have a significantly higher clock speed and CPU cache. I wasn't expecting lowered CPU usage on four cores though. Looks like Ryzen is the next frontier for media streaming.

dahartigan

@poisson said:

@dahartigan said:
An update for anyone interested. I have moved my Plex over to the Ryzen L to see how it performs and I have noticed that things are much faster, especially with transcoding. The CPU usage is actually significantly lower than when run on Intel processors, as is the system load.

I was shocked to see that without any change to the usage pattern, this seriously outperforms everything else this has run on. The crazy part is my 24 hour CPU usage average is 9% for 4 threads of Ryzen whereas previously it was 15% on 12 EPYC threads with the same load.

I think I'll continue this experiment for a month and update my findings in a week, perhaps I need to work on a Plex benchmark to somehow quantify this.

The Ryzens have a significantly higher clock speed and CPU cache. I wasn't expecting lowered CPU usage on four cores though. Looks like Ryzen is the next frontier for media streaming.

It was a complete shock to me, but I will see over the next week how it goes. I'm super conscious of my CPU usage so if it starts to get closer to 25% I'll panic-move it back.

cybertech

It certainly won't be long before a 4 core vps can run Plex transcodes comfortably

gtxr

I hope I can win the new year’s prize

samm

@gtxr said: I hope I can win the new year’s prize

did you purchase anything ? I think they will only upgrade resource when you buy VPS, no feebie.

gtxr

Yes, I grab Ryzen L Plan.
The cpu ia Beast!

@samm said:

@gtxr said: I hope I can win the new year’s prize

did you purchase anything ? I think they will only upgrade resource when you buy VPS, no feebie.

poisson

@gtxr said:
Yes, I grab Ryzen L Plan.
The cpu ia Beast!

@samm said:

@gtxr said: I hope I can win the new year’s prize

did you purchase anything ? I think they will only upgrade resource when you buy VPS, no feebie.

NICE

gtxr

New bench!

---------------------------------------------------------------------------
 Region: Global  https://bench.monster v.1.4.9 2019-12-24
 Usage : curl -LsO bench.monster/speedtest.sh; bash speedtest.sh -Global
---------------------------------------------------------------------------
 OS           : CentOS 7.7.1908 (64 Bit)
 Virt/Kernel  : KVM / 3.10.0-1062.1.1.el7.x86_64
 CPU Model    : AMD Ryzen 9 3950X 16-Core Processor
 CPU Cores    : 5 @ 3493.434 MHz x86_64 512 KB Cache
 CPU Flags    : AES-NI Enabled & VM-x/AMD-V Enabled
 Load Average : 0.20, 0.34, 0.21
 Total Space  : 50G (17G ~35% used)
 Total RAM    : 4796 MB (495 MB + 3271 MB Buff in use)
 Total SWAP   : 4095 MB (0 MB in use)
 Uptime       : 0 days 0:8
---------------------------------------------------------------------------
 ASN & ISP    : AS18450, WebNX, Inc.
 Organization : WebNX, Inc.
 Location     : Los Angeles, United States / US
 Region       : California
---------------------------------------------------------------------------

 ## Geekbench v4 CPU Benchmark:

  Single Core : 5869  (EXCELLENT)
   Multi Core : 21577

 ## IO Test

 CPU Speed:
    bzip2     : 144 MB/s
   sha256     :   1.0 GB/s
   md5sum     : 582 MB/s

 RAM Speed:
   Avg. write : 3822.9 MB/s
   Avg. read  : 10820.3 MB/s

 Disk Speed:
   1st run    : 1.0 GB/s
   2nd run    : 1.0 GB/s
   3rd run    : 1.0 GB/s
   -----------------------
   Average    : 1024.0 MB/s

 ## Global Speedtest

 Location                       Upload           Download         Ping
---------------------------------------------------------------------------
 Speedtest.net                  891.35 Mbit/s    932.28 Mbit/s    9.682 ms
 USA, New York (AT&T)           175.31 Mbit/s    271.42 Mbit/s    75.598 ms
 USA, Chicago (Windstream)      52.21 Mbit/s     321.16 Mbit/s    39.464 ms
 USA, Dallas (Frontier)         310.96 Mbit/s    590.25 Mbit/s    44.478 ms
 USA, Miami (Frontier)          117.47 Mbit/s    317.73 Mbit/s    73.517 ms
 USA, Los Angeles (Spectrum)    462.51 Mbit/s    379.09 Mbit/s    22.750 ms
 UK, London (Community Fibre)   140.15 Mbit/s    117.83 Mbit/s   119.146 ms
 France, Lyon (SFR)             107.02 Mbit/s    84.38 Mbit/s    137.542 ms
 Germany, Berlin (DNS:NET)      114.56 Mbit/s    196.31 Mbit/s   159.143 ms
 Spain, Madrid (MasMovil)       109.65 Mbit/s    253.86 Mbit/s   140.390 ms
 Italy, Rome (Unidata)          87.74 Mbit/s     58.01 Mbit/s    159.545 ms
 Russia, Moscow (MTS)           71.80 Mbit/s     152.54 Mbit/s   196.963 ms
 Israel, Haifa (013Netvision)   17.45 Mbit/s     167.89 Mbit/s   189.619 ms
 India, New Delhi (GIGATEL)     22.60 Mbit/s     88.41 Mbit/s    272.749 ms
 Singapore (FirstMedia)         19.66 Mbit/s     14.13 Mbit/s    231.684 ms
 Japan, Tsukuba (SoftEther)     133.93 Mbit/s    106.57 Mbit/s   104.292 ms
 Australia, Sydney (Yes Optus)  55.15 Mbit/s     59.69 Mbit/s    166.591 ms
 RSA, Randburg (Cool Ideas)     19.04 Mbit/s     28.86 Mbit/s    290.636 ms
 Brazil, Sao Paulo (Criare)     50.21 Mbit/s     141.73 Mbit/s   164.991 ms
---------------------------------------------------------------------------

 Finished in : 9 min 45 sec
 Timestamp   : 2020-01-04 02:40:32 GMT
 Saved in    : /root/speedtest.log

 Share results:
 - http://www.speedtest.net/result/8913145217.png
 - https://browser.geekbench.com/v4/cpu/15100345
 - https://clbin.com/XmxXK

dahartigan

I'm extremely happy with my Banditos, the EPYC and the Ryzen are working very smoothly - not even breaking a sweat - dragging the plow along with conviction.

If I had to choose which one to keep, I wouldn't be able to decide. I think I will keep both just to avoid having to choose.

Delong

@dahartigan said:
I'm extremely happy with my Banditos, the EPYC and the Ryzen are working very smoothly - not even breaking a sweat - dragging the plow along with conviction.

If I had to choose which one to keep, I wouldn't be able to decide. I think I will keep both just to avoid having to choose.

Haha. Man, I just decided to hop on to take a peek at what's up RIGHT as you commented. Are you stalking me lol.

Sorry to put you in that spot man, can't help but offer EPYC VPSes(s?) to keep things RYZEN. A for effort right?

sonic

@dahartigan said:
If I had to choose which one to keep, I wouldn't be able to decide. I think I will keep both just to avoid having to choose.

Such a hard decision!

dahartigan

@Delong said:

@dahartigan said:
I'm extremely happy with my Banditos, the EPYC and the Ryzen are working very smoothly - not even breaking a sweat - dragging the plow along with conviction.

If I had to choose which one to keep, I wouldn't be able to decide. I think I will keep both just to avoid having to choose.

Haha. Man, I just decided to hop on to take a peek at what's up RIGHT as you commented. Are you stalking me lol.

Sorry to put you in that spot man, can't help but offer EPYC VPSes(s?) to keep things RYZEN. A for effort right?

Haha that's hilarious :-) I'm always stalking and prowling forums, bumpin' threads, droppin' benches, kickin' my 2 cents in, ya know? :-)

You actually get an A+ for effort, all jokes aside, I'm very impressed with the service and performance and I'm sure others would definitely agree with me on that.

@sonic said:

@dahartigan said:
If I had to choose which one to keep, I wouldn't be able to decide. I think I will keep both just to avoid having to choose.

Such a hard decision!

It's probably easier to calculate the exact number of potassium atoms in the universe than it is to decide which of these to keep.

My preferred approach is to keep both and assume the universe is one giant potassium atom.