I used prices from US as Anandtech is primarily used by people from USA.
Don't go off-topic... :awe:
It was about FX 8350 vs Xeon 5650 on stock clocks and Xeon 5650 would lose in majority of benchmarks and specially in case you want to play games on it. StarCraft 2 would ran 30% faster on FX 8350 than Xeon 5650 which was being discussed.
Anyway you can only buy used i7 990X by now and you'd be lucky if you could buy unused one or even a retailer that still sells it so you can also have a warranty for it. :hmm:
That's interesting. I didn't know that WCG existed before BOINC. I thought that it started because of BOINC.
I'm not a noob to DC either, I started with distributed.net's RC4-56 challenge, with (IIRC) a Pentium MMX 166 @ 233. At least, I think that they were able to leverage MMX back then somehow. Memory of those days is a little hazy.
I also used to run SeventeenOrBust for a long, long time, using their dedicated client.
I also ran Prime95 for a while too.
My participation in WCG is more rather recent, however.
What are you talking about? There's people on WCG that have over 600 years of CPU time. I know you're a newbie to WCG, but some of us spent many years crunching. I was crunching for years before they used BOINC.
Oh, and you have no clue how much processing power I have access to.
Now that you have managed to completely derail this thread I will not reply to you any further.
WCG was started in 2004 and initially ran with United Devices software. In 2007 they switched to BOINC.
http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=15715#122294
Those of us that were running UD were at some point supposed to get additional credit - I think it was 7:1 - due to points being calculated differently between the two systems. That never happened though.
+50%? Do you have anything that can demonstrate this? I typically see ~20-25% for rendering.
Move south, FX's are cheaper.
@MiddleOfTheRoad, how you put it is confusing though. Number of projects is not relevant, because a quad core Q6600 would turn in 4 projects in the same time a dual core Haswell Pentium turns in 4 - because the Pentium is turning them in twice as fast.
If an i5 has ~60% better IPC than an FX, and both are clocked the same, the i5 will still only turn in 6.5 projects for every 8 an FX turns in, because the FX has more total throughput. However, if an i7 is gaining another 50% from hyperthreading, it will be turning in 9.75 for every 8 an FX turns in.
The problem with the FX is not its total throughput - it's still a respectably fast chip, especially considering its price - but with how it's distributed across its cores. An i7 effectively has 4 very fast cores, 60%+ faster each than an FX core, and 4 slower "cores" from HT. This is a better way of distributing things because with the same total throughput you have significantly better performance in loads that can't scale well or distribute across a lot of cores, which is nearly all loads non-scientific in nature.
he was probably mistakenly talking about non OoO archictectures. Cell, Xenon, Atom
^ This.
I was careful to word my post, but didn't feel it was necessary to state that I meant IPC was on a core to core basis, given how frequently it's discussed in that context on this forum. I understand how the math works. I assumed that "60% better instructions per clock (per core)" would be understood as a rate of completed instructions, or more clearly, "over a given period of time, an i5 will do 60% more, per core".
Basic math shows that an FX has more total throughput than an i5 across all of its cores, even with a modest module penalty, if each of an i5's cores is only ~60% faster.
(4 * 1.6) < (8 * 0.9)
it's hotter down south, so they have to incentivise the purchase
Would it make sense to use one for a server build?
Would it make sense to use one for a server build?
^ This.
I was careful to word my post, but didn't feel it was necessary to state that I meant IPC was on a core to core basis, given how frequently it's discussed in that context on this forum. I understand how the math works. I assumed that "60% better instructions per clock (per core)" would be understood as a rate of completed instructions, or more clearly, "over a given period of time, an i5 will do 60% more, per core".
Basic math shows that an FX has more total throughput than an i5 across all of its cores, even with a modest module penalty, if each of an i5's cores is only ~60% faster.
(4 * 1.6) < (8 * 0.9)
No, it doesn't. Funny scientific calculation being used as a case to justify acquiring an FX, because this is exactly *the* case where the FX doesn't make sense, as whatever savings you make when acquiring the chip gets lost by the extra energy bill the server will get. This is the reason AMD still has a presence in gaming desktop but is heading towards 0 on the server and workstation market.
The only companies that go Opteron for servers I know of are AMD and Globalfoundries, both for obvious reasons.
The answer is both, we are using the Intel Xeon class processors and a bunch of our infrastructure is using the Intel. Recently we switched to AMD Opterons, and as you wonder why, one of the things we are looking for is increasing the memory per host and the Opterons allow us in a single socket configuration to address more memory, and so all of the new deployments we are putting out there are carrying 64GB per host and the 8-core Opteron processors.
The AMD Opteron 6338P processors are designed to meet todays enterprise workloads, making 1&1 Internets AMD powered XL servers ideal for dynamically generated websites, xSQL databases, CRM and storage. To support these heavy duty, compute and memory intensive workloads, 1&1 Internets customers will be able to take advantage of key features in the AMD Opteron 6338P processor, including:
- 12 64-bit x86 cores using our Piledriver architecture, a tried and tested architecture that designed for today's multi-threaded workloads
- AMD Turbo CORE technologies boost each core from 2.3GHz to 2.8GHz, providing a power efficient performance boost when it is really needed
- AMD Virtualization technology, to provide superior performance for virtual machines
- AMD-P, our state-of-the-art power management technology, enabling systems using the AMD Opteron 6338P processor to be extremely power efficient
That, and they don't salt the roads anywhere near as much, so your FX won't . . . er, wait . . . nevermind.
For a home server, yes. They're also quite handy for VM work. If you're looking at serious professional work in an office/corporate environment, it's difficult to justify large numbers of AMD machines (TCO and all that). I expect there are a few situations where they make sense. Probably around 2.5%-4.5% of the time, judging by AMD's share of the server market.
The newer "E" versions are definitely a great choice for a low cost home server. DrMrLordX is spot-on -- the built-in hardware virtualization of the FX is very strong.
AMD-P, our state-of-the-art power management technology, enabling systems using the AMD Opteron 6338P processor to be extremely power efficient
Yeah, I know that if we look hard enough we'll still see an Opteron product here and there, but far from meaningful, in the end it doesn't change the fact that AMD has been basically wiped out on the server business.
Saying far from meaningful is really an out of touch from reality statement, deal with Verizon is very meaningful and 1&1 is potentially too...
AMDs SeaMicro server product family currently supports the next generation AMD Opteron (Piledriver) processor, Intel® Xeon® E3-1260L (Sandy Bridge) and E3-1265Lv2 (Ivy Bridge) and Intel® Atom N570 processors. The SeaMicro SM15000 server also supports the Freedom Fabric Storage products, enabling a single system to connect with more than five petabytes of storage capacity in two racks. This approach delivers the benefits of expensive and complex solutions such as network attached storage (NAS) and storage area networking (SAN) with the simplicity and low cost of direct attached storage.
AMD, which acquired SeaMicro last year, has been touting this as a huge victory over Intel and its dominance in the server market. We can exclusively reveal, however, that more than three quarters of the SeaMicro servers purchased by Verizon are actually powered by the Intel Xeon E3, not AMDs own Opteron chip. AMD has, rather ironically, become an Intel OEM.
According to an industry source with first-hand knowledge of Verizons cloud computeimplementation, who spoke to ExtremeTech on condition of anonymity, The vast majority of the CPUs (more than three quarters) AMD is using for this implementation with Verizon are Intel Xeon E3s using the SeaMicro fabric they acquired. So in essence AMD is now becoming a microserver OEM for Intel and is still largely selling Intel parts in microservers for their customers.
That, and they don't salt the roads anywhere near as much, so your FX won't . . . er, wait . . . nevermind.