New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Phoronix Test Suite Apache benchmark gives 2730.21 http://openbenchmarking.org/result/1704271-RI-ARMV82GAP69
Does anyone know the results for the C1?
Is installing windows on these possible?
I don't think so. Right now, only Ubuntu Xenial and Debian Jessie are available on these.
I'm a bit lost right now. What would be the best choice of scaleway 2.99 if I want to run ubuntu and lets say plex media server (with fuse/sshfs drive) and some other stuff? So there is armv8 vs X86-64 right now, are you benchmarking these against eachother?
And C1 BareMetal for 2,99 is also something different?
I have a kidechire and these new Scaleway ARMs feel a bit faster. I'm not a big fan of Scaleway but I will most likely be keeping one of these new ARM servers around especially since they're cheap.
The C1 is the predecessor of these new ThunderX-based ARM servers. The Starter Cloud and C2 are Atom based and outperforrm the ThunderX, so these are probably what you'd want. I don't think Plex supports ARM anyway.
Ah I thought as long as ubuntu is running the CPU architecture doesn't matter, but that argument also doesn't work for 32 vs 64bit...
The x86 version is same price, same other specs, but the CPU is 2x faster. What is your reason to keep ARM?
Most software is only made available for x86 (which is the name of the architecture, not 32/64 bit), and a x86 binary does not run on ARM. Since Linux distributions are made from open source software, they are able to build for ARM and other platforms, most software in the repositories will therefore work. But with proprietary software, you're entirely dependent on the vendor to provide ARM binaries.
Thank you for the explanation!
Windows doesn't run on ARM.
Windows RT, tho..
Well I was going to say the dedicated cores...but then I went back over their pricing page and it looks like these are just VPSes with ARM cores and not like the C1s? I opened a support ticket asking if they were dedicated cores...and their support basically said they don't know even know and just copy & pasted the pricing =/.
Shared core ARM. That's like when your sister grabs your tablet and starts texting when you're trying to get your YouTube on.
I spun up an 8 core one and compiled ffmpeg from source snapshot in 5m 44s on 8 threads. IIRC the 4-core C1 took around 20m so the 64 bit cores are around 2x as fast. That puts them comparable to the C2750 Atom cores I'd say. Openssl speed rsa2048 -multi 8 did 510 private key operations per second which is unimpressive, but the code may be poorly optimized for this platform compared with the x86. A few other informal tests were in the same ballpark shown by the ffmpeg compile.
@willie - A compile isn't really that great- it shows you how fast it can turn text files into code. That's about 0% of most VPS uses, unless you are too lazy to setup a cross-compiler. I'd like to see how the C1 and it manage to re-encode a sample mkv with both audio and video containers and codecs changed.
I did a transcode on a C1 a while back and its speed was about in line with other tests, or maybe slower. I've spun down the arm64 but would expect similar. What you get from it is lots of slow cores, so unless you're trying to transcode lots of files in parallel this isn't a good machine for the purpose.
Compiling is quite cpu intensive and I do it on vps sometime, though usually on a dedi.
It's generally briefly intensive, though- and doesn't give you a good idea of how well the entire platform handles under a semi-consistant load unless you have a slow-as-shit system, or are building world.
That said, I should see if my core2 ever finished making the mips cross compiler I started last night.
I find compiling is a representative workload for the stuff I do, though I haven't compared transcode speed on many machines carefully. I figure if if someone really wants to transcode a lot, a GPU beats all alternatives.
I tend to use my fastest machines and cross-compile unless they're using different libraries and platforms.
Scaleway IPv4 NAT is odd.
I mean i get it, but(t) i can live without it.
How so?
Compiling ffmpeg takes around 3 minutes on an i7-3770 on all 4 cores, so I figure that's a fairly steady load. I compile it pretty often since I pull from git regularly. Compiling gcc takes 30 minutes or so but I've only done it a few times with recent versions.
Windows RT is a scam.
Please explain?
IPv4 is slowly going away, becoming more rare and more expensive. Meanwhile IPv6 is free, and it becomes more popular, as ~20% of Google searches already come from IPv6. We can slowly live without IPv4 and focus on IPv6.
But(t) a big entry like Scaleway does not like IPv6 to invest in it, they clearly invest in more servers connected with already consumed public IPv4 behind NAT (IPv4 is reserved and can be swapped between servers; all servers get private ip 10.x.x.x). Plus, they charge you for it: 1 Euro, because IPv4 is rare and we should understand this business model.
For the sake of us asking, they added a free /128 IPv6 (not /48 nor /64) only for some servers, and that /128 IPv6 is random with every server restart (different IPv6 when you start/stop server).
I think some of the NAS versions of PLEX are ARM based so it might be possible