28 posts • joined Friday 26th October 2007 14:23 GMT
Reason to upgrade...
Power consumption anyone?
I upgraded from a Athlon64 X2 4200 last year to a Core i5 3550. The old system used to idle at 100W (measured at the wall). The new one is less than 50W.
Newer machines won't be quite that dramatic a difference, but there are still some savings that can be made by replacing older desktops.
Re: Simple solution
Except the thieves are too stupid to tell fibre and copper apart. I've seen several cases of BT fibre infrastructure being ripped out by thieves who thought it was copper.
Re: Not to throw a damper on this lovefest...
Do you have a source you can cite for that? All the articles I've seen state Haswell supports OpenGL 3.2 and OpenCL 1.2.
Re: take a leaf from ARM
Isn't the process migration something the OS should be doing? AMD and Intel already shut down many bits of their CPUs (including whole cores) when they're not in use.
AMD are already doing that with their latest generation graphics cards, they're calling it 'ZeroCore'.
The problem with lots of small servers is the overhead. Even your dinky little ARM servers are going to be using some power all the time. Whereas in a virtual environment you'll have fewer servers so the base load of them sitting idle will be less. You can even shut down servers completely if the load isn't sufficient for them all to be running.
That was presumably the E5-2600 version of the DL360 Gen8 (which appear to have gained a p suffix), whereas these are DL360e Gen8 (note the different suffix).
Nice way to confuse people HP. At least Dell are giving their systems different numbers (their equivalents are R420 (E5-2400) and R620 (E5-2600).
Re: Intel thoroughly missing the point here
I don't think Intel are missing the point, they just see gaps in their line-up that need filling. (i.e. between single socket and a full-on dual socket, and between dual socket and their big E7 quad setup).
Intel cores are currently faster, clock for clock. For example, lets take anandtech's review of the E5-2600 ( http://www.anandtech.com/show/5553/the-xeon-e52600-dual-sandybridge-for-servers/ ). In that they take a simulated E5-2630 (6 cores, 2.3GHz) and they reckon it beats an Opteron 6276. Thats a $616 CPU beating a $788 CPU (list prices for both).
But still, the Opteron reaches futher down than the 2600 series does, so Intel knocked off a few of the extras to produce the 2400 series to compete with those. Likewise, the E7 is overkill for many where you just want 4 sockets-worth of ram and cpu, so bump up the E5 to fit there.
Really a fix?
I may be wrong, but I was under the impression that these things are essentially still using IE6, but hiding it round the back. What happens when XP support goes? You're still stuck using components that MS aren't supporting, so you'll be stuffed in just the same way as soon as a bug comes up.
Indeed. They've typoed the name (It's a 6904 card, not a 6094). And the 44 ports is for a full 6913 chassis. It's 4 ports per card.
http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps708/data_sheet_c78-696623.html if you're into your hardware porn.
They're only following in Apple's steps. You had noticed that as soon as Lion released, Apple stopped bothering with any more updates for Leopard? The last PPC-compatible version...
There's a paper http://www.thoughtcrime.org/papers/ocsp-attack.pdf on how the protocol can be MITM attacked.
Rather than bandwidth usage, resource usage of the server running the OCSP service might be an issue if you get millions of requests. If it takes a while to respond, people will complain their browsing is slow (you're waiting for OCSP to respond before trusting the site). If it fails to respond at all, do you block access to the site? (That'll be disabled if people want to get to their stuff). Or allow access, risking people going to a site that shouldn't be trusted?
Re: Gun to the head
Are you actually maxing out the ram on all your hosts? As long as your VMs are collectively using less than 32GB x number of CPU sockets, your licences are good. If you're just over, you might be better buying a few additional Enterprise to get additional licences than upgrading to Enterprise Plus
Sandy Bridge gaming
The problem for AMD, is the pricing of the new kit appears to be higher than for a basic Core i3. With the difference, you can buy yourself a nice discrete card as good as the integrated unit in AMD's APU.
So, a faster processor and similar graphics for the same money. Only if you absolutely have to have it integrated does AMD look a winner.
Re: AMD has them beat for Virtual Machines
Is that still the case? I know HT used to be a bit of a dog, but the current versions are much better.
VMware's vSphere best practices guide states "For the best performance we recommend that you enable hyper-threading".
Anandtech have some benchmarks from when Opteron 6100 and the Xeon 7500 were new and the Xeon was faster ( http://www.anandtech.com/show/3894/server-clash-dellr815/10 ).
If the E7 is faster still...
Not Richard Desmond
It's the Express and the Star that are his papers.
Doesn't stop them being rubbish though :)
The suggestion of the article is that most of these providers are only 'happy' for you to download that much because most of their customers don't.
If everyone was maxing out their connections they might have to charge something closer to what BT Wholesale want.
Re: Imitation is the most sincere form of flattery
If you put it into the server chips, you start to make it more attractive to do your load balancing with nice cheap servers, rather than buying expensive black boxes from the aforementioned vendors.
If you want it on the desktop, buy a Core i5 (or a i7-980X).
Via may have had acceleration for a while, but Intel chips were probably still faster even without.
The 8-core chips from Intel aren't entirely comparable. (aka Beckton, Nehalem-EX or Xeon 7500)
They're more akin to being a pair of 5500s in a single package. So they don't have all the architectural advantages of the 5600s.The other main difference is they use different sockets. 5600s go into boards that only take 2 processors, Beckton will go into 4-way and above.
Mac Pros use dual-socket Xeon 5500s, rather than the single-socket Core i7 line.
So the upgrade would be likely to be the dual-socket Xeon 56xx range rather than a i7-980X. Going from a pair of quad-cores to a single hex-core would normally be considered a downgrade.
This story is about sites that have been coded with all the crappy hacks needed for IE6, but are just feeding IE8 that version rather than a properly standards compliant page that IE8 can handle. IE8 then has to be told to pretend to be IE6 to render 'properly'.
I don't disagree the treaty is a load of cobblers.
But making factual errors weakens any argument. As the article now says the treaty is one-sided is rather different from the original version saying the US haven't been bothered to ratify the thing.
That was why there was Fail.
The treaty was ratified by the US several years ago - http://press.homeoffice.gov.uk/press-releases/UKUS-extradition-traety?version=1 (nice bit of spelling by the homeoffice there)
But if you visit the US then you're bound by their laws. Generally you're under the law of the county where your actions happen. The tricky bit with computers is defining where the action is. The servers are in the US, so it's not unreasonable to consider whether the crime was committed in that country.
- IT bloke publishes comprehensive maps of CALL CENTRE menu HELL
- Nine-year-old Opportunity Mars rover sets NASA distance record
- Prankster 'Superhero' takes on robot traffic warden AND WINS
- Analysis Who is the mystery sixth member of LulzSec?
- Comment Congress: It's not the Glass that's scary - It's the GOOGLE