I always highly discourage people from using their debit cards online. Why?
If your debit card is compromised a lot of very bad things can happen to you even if you eventually get all the fraudulent charges refunded. For example in between the time the card is compromised and the time you figure it out, you could be bouncing checks all over town. And after that its up to you to fix your credit rating and convince your land lord or mortgage company it wasn't your fault
I never use anything but a credit card online because my liability is limited and it can't backfire into other areas of my finances like the debit card would. And credit card != pay later. There is nothing stopping me from buying now and paying off at the next statement
Nehalem was pretty power hungry. Efficiency under load *might* have been similar to a modern FX chip (unlikely), but their idle draw was much, much higher. There's no arguing that an overclocked 6 core Nehalem isn't faster than an FX-8 though.
EDIT: Is -> Isn't
That is pure garbage coming from someone who clearly has never run any scientific computing. You can't change the massive architecture disadvantage of only running 4 projects simultaneously versus a CPU that can run 8.
The FX 8320e is a far superior chip to your i5 for those running multithreaded scientific calculations. Your original quote was "In short you need an aggressively clocked AMD FX 8 core chip to even come somewhat close to a stock clock core i5 Devils Canyon in most tasks." That is clearly incorrect -- a stock clocked FX octacore will soundly thump any i5 on the World Community Grid any other grid computing projects where integer performance matters. That faster IPC on only 4 threads can't keep up with running 8 projects simultaneously. I've run both chips personally -- the i5 is considerably slower.
My i7 4790K is a monster on BOINC that can run 8 threads faster than any FX can, but you were talking about the i5.
If FX 8350 was also overclocked then it would beat overclocked 6 core Nehalem.
Car enthusiasts laugh at this "expensive hobby".
@ tential: Exactly. Or one month's internet, or one month's cell phone bill, or a few drinks or going out to eat a nice dinner. And people dont even need to start with the insults that I think everyone needs quad Titans and a hex core intel. That is certainly not the case. OTOH, PC gaming is an expensive hobby, and picking an inferior (for gaming) cpu that uses more power to boot in order to save 10% or less of the cost of a system just seems like false economy.
You've posted this falsehood in the past and I debunked it then. Here's where http://forums.anandtech.com/showpost.php?p=36786766&postcount=119
Xeon 5650 6 core Nehalem idle; 143 watts
FX 8350 8 core Piledriver idle; 95 watts
Both are 32nm, 5650 die size is 240mm^2 compared to FX 8350's 320mm^2.
If FX 8350 was also overclocked then it would beat overclocked 6 core Nehalem.
No it wouldn't. Those Xeons are only like ~10% slower than Sandy Bridge-E at the same clock and OC like champions. A stock Core i7 990X beats the FX9590 and any of those Xeon chips can reach much higher clocks than a stock Core i7 990X.
my tests with Blender show similar results between FX and Intel's offerings.
Like what? Rendering and transcoding are not it.
http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/10
The 3770K does better in Cinebench 11.5 only due to this soft being Intel optimised and giving this latter a 15% advantage, same thing in Euler 3D wich is exclusively Intel optimised, (these are FPs benches) that is, only with exclusives optimisations Intel CPUs show superior perfs...
my tests with Blender show similar results between FX and Intel's offerings.
There are some weird results in there. Do they not account for overclocked and just list the original CPU regardless of operating speed or something? There's lower clocked i5's showing way higher scores than higher clocked, same family, for example. Hard to sort and navigate.
You've posted this falsehood in the past and I debunked it then. Here's where http://forums.anandtech.com/showpost.php?p=36786766&postcount=119
can you explain what you mean more? similar results to the guy I quoted?
There are some weird results in there. Do they not account for overclocked and just list the original CPU regardless of operating speed or something? There's lower clocked i5's showing way higher scores than higher clocked, same family, for example. Hard to sort and navigate.
Seriously, even engineering samples are listed in there -- I can't see how anything of tangible value can be determined from those charts. How do you evaluate the performance of engineering samples? It is in their nature to have bugs and performance glitches.
@MoToR:
I dont understand what you are saying either. How can an i5 be a bad as you make it sound, and an i7 so much better. The absolute maximum increase I have seen in any benchmark is around 30% due to hyperthreading. So if the order of power is i5<FX<i7,
how much faster is the i7 vs the i5? If there is a total spread of 30% between the i5 and i7, based on general best case scenario for hyperthreading, how can the FX be so much better than the i5. Maybe I am all wrong here, I admit I am not familiar with your workload, but based on everything I have seen about i5 vs i7 what you are saying does not make sense to me.
Depends on the code. If you are using a lot of expensive functions in mathematical code (log, sin, cos, sqrt, division, etc), HT can bring massive gains when the pipeline stalls. +50% would be common.
For the record I want to add that the extra $1-$2 a month I might pay in electric is a perfectly valid point. I'm not arguing that. I'm just arguing how that really effects someone's budget vs. having to pay more up front. It is worth bringing up, I'm not trying to discredit that. But, I think an extra dollar or so in power costs a month is a lot easier pill to swallow than someone on a thin budget paying as much or more for an i5 CPU as you would have to pay to get an FX8310 + motherboard + cooler.
It depends on your budget, and I also agree that saving an extra $50-$100 for a better system can be worth it. But, I still also believe the FX's performance is generally more than acceptable, as is an i5's performance. In practical real world difference in performance, not numbers but how well you feel the performance difference, I'm not sure the gap between the two camps (in that price range) is really as much as you might think.
Xeon 5650 6 core Nehalem idle; 143 watts
FX 8350 8 core Piledriver idle; 95 watts
Both are 32nm, 5650 die size is 240mm^2 compared to FX 8350's 320mm^2.
If FX 8350 was also overclocked then it would beat overclocked 6 core Nehalem.
+50%? Do you have anything that can demonstrate this? I typically see ~20-25% for rendering
+50%? Do you have anything that can demonstrate this? I typically see ~20-25% for rendering.
First off, your memory sucks. I never compared the FX to an i5 previously, I was discussing i7 3770k versus FX 8350 (as compared in a Phoronix linux review). And a 4 threaded i5 is a completely different animal from an 8 threaded i7 in regards to scientific computing.
Second, you might have debunked it in your fantasy world. But in reality, the FX-8320 was/is considerably faster than our Ivy i5 which is why we upgraded it to the i7 4790k. The FX is no match for Haswell 8+ thread chips (and I never said that), but an octocore FX is generally faster than the vast majority of Intel 4 threaded chips on the grid. If someone can do a "Super" overclock on an Intel, it's probably a possibility -- but in general terms, more threads = more projects completed for BOINC. The bottom line is teams get points for completed projects, even it takes 20% longer to complete a project -- an 8 thread chip is processing twice the projects of a 4 thread chip simultaneously. This is not rocket science.
Our team has 3.5 million points -- and the workhorse of our desktops is our FX-8320. The i5 couldn't post worthwhile daily numbers to keep it in our mix. Our Haswell i7 is a monster, though.... The FX is clearly 2 years older running next to our brand new i7.