What is Digg?21 Comments
- smb3d, on 10/12/2007, -0/+6Why on earth would they do that?
- Lazybones, on 10/12/2007, -1/+6Not it is comparing what they have on the market, NOW that you can purchase, not those systems that or in the not so distant future that will always be just around the corner.
- nofxjunkee, on 10/12/2007, -0/+4New Xeons are based on the same technology as the Core processors. Sossaman is a derivative of Yonah with enhancements to support dual processor boxes and other things.
More reading:
http://en.wikipedia.org/wiki/Intel_Core#Sossaman
http://en.wikipedia.org/wiki/Xeon#LV.2C_Core_Duo-based_.22Sossaman.22 - nofxjunkee, on 10/12/2007, -0/+3Sorry, I linked to the wrong information. The Xeon 5100 (Woodcrest) is based on Core 2 (Merom and Conroe), which is based on Core (Yonah).
http://en.wikipedia.org/wiki/Intel_Core_2
http://en.wikipedia.org/wiki/Xeon#5100-series_.22Woodcrest.22 - TonyCubed, on 10/12/2007, -0/+3Thats the original Core Duo, not the new one.
- HappyScrappy, on 10/12/2007, -0/+3Why argue about it in the theoretical?
Anand already tested the AMD 65nm chips. The results good, but not fantastic.
http://anandtech.com/cpuchipsets/showdoc.aspx?i=2889 - geminitojanus, on 10/12/2007, -1/+4Hate to burst your bubble, especially after you started throwing around the "fanboy" term, but the 65nm shrink of the AMD processors isn't initially going to help them perform better; it'll give them a boost of maybe 5-10% in performance, but a very nice drop in heat (which will allow them to start ramping performance again, hince the reason for the shrink).
AMD has a very wide deficit to recover here, in many cases upwards of 50% (and in at least two benchmarks, 100%), which is architecturally impossible to overcome without adding the fourth execution unit and decreasing the latency of their SSE implementation to single-cycle (all of which will be included in Barcelona, which is expected H207). The fact just is: at this moment in time, Intel has the best processor on the market, and they're using this fact to market the hell out of it.
So, let's be mature about it, and read the numbers and discuss what they mean, what the companies are doing right in their implementations, what they're doing wrong, and not how we should cheer on either company (as both companies working to better their products mean that we get better products for cheaper, which is an all-around WIN). - geminitojanus, on 10/12/2007, -0/+2This has been the case for a while now, as the AMD64 architecture revolves around the integrated memory controller, which cuts down quite a bit on DRAM-to-L2 latency. While this is a great innovation, it doesn't mean a whole lot in the end if your processor isn't doing enough work to keep up with its magnificent memory management.
It'd be very interesting to see what difference it would make if AMD removed the memory controller from their chips, or if Intel added one to theirs, to see how much latency actually plays into current workloads. If I had to guess, I would say not a lot except for in certain gaming situations, but I've been surprised before on benchmarks like this.. - Complexium, on 10/12/2007, -1/+3Funny to see how the Opteron-box smashed Intel on the SiSoft Sandra memory bandwidth benchmark.
- HappyScrappy, on 10/12/2007, -1/+2The test is valid, but to be honest, FB-DIMMs are holding back the Xeon. You'd use a lot less power and have a lot less latency (less than the AMD) if you used a system with DDR2.
If you need a lot of RAM, FB-DIMMs are fine. But everyone else should use DDR2 to save money up front, improve performance and save money spent on power over the lifetime of the product.
And that includes you, Apple. FB-DIMMs really hurt the Mac Pro for all but the highest-end customers. Make a more sensible alternative for the rest of us, please. - falloutsyndrome, on 10/12/2007, -1/+2I'm glad to see that there is some life left in AMD, considering their processors have nearly the same performance, but are still using a 90nm process, i can't wait to see what their 65nm dishes out. I'm glad I didn't jump out and buy an core 2 chip just yet.
- rickjames47, on 10/12/2007, -0/+1Maybe you're thinking about the Itaniums ... Dell and HP have ditched them in their servers and Intel has cut production of about 8 of them so ...
- markdde, on 10/12/2007, -0/+0this makes it harder to choose a new server.
While the benchmarks look promising, AMD had always offered far better performance in the past, especially in Mysql. I think I'll wait to see what AMD has to return before switching. - thejackamo, on 10/12/2007, -2/+2I personally don't like how the test in convoluted by having both the 5100 and 5300 - I thought this test was supposed to just compare the dual core offerings. Not to say that I'm not interested in seeing how the numbers stack up, but it doesn't seem germane to the test at hand - you should test dual-core to dual-core and quad to quad and not some kind of mixed-processor mashup.
- TonyCubed, on 10/12/2007, -3/+2The latest Xeon's aren't based on Core 2 duo are they though?
- stmiller, on 10/12/2007, -5/+4Yes, I realize that. But this is EXACTLY like the ATI/Nvidia 'who has the top-card now' war that goes on throughout the year. Comparing different generations of technology to one another. In this case the Xeon SHOULD have better marks b/c it is a 65nm chip.
It's crazy how Intel suddenly has a huge fan boy crowd, when techies before always preferred AMD. (For example I am dug down, and you are dug up.) I guess that is how trends go. - inactive, on 10/12/2007, -3/+2 " wouldn't work for an apples-to-apples comparison "
INTEL IS PAYING OFF PEOPLE FOR THEIR CHIPS
READ FIRST
Talk ***** later
just look at the FSB people deadlock occurs there
ps: I have a Intel core 2 duo and my AMD FX box is faster ! - stmiller, on 10/12/2007, -4/+2That AMD chip still isn't 65nm. Sort of like comparing the latest ATI or Nvidia card (whoever has the current lead) vs. the competitors previous generation card.
I would hope that the Xeon would be faster. - jstone, on 10/12/2007, -5/+3@stmiller:
I would assume that the Intel fanboy crowd (at least around digg) has something to do with Apple using Intels now. Many hard-core apple fanboys will blindly support anything Apple does/says/uses. - Unbekannt, on 10/12/2007, -4/+0Nice work Scott! One day I need to show you how to superimpose the latency graphs :)
- dogman, on 10/12/2007, -13/+1I thought Intel was discontinuing Xeons?
Digg is coming to a city (and computer) near you! Check out all the details on our 
© Digg Inc. 2009
— Content created and posted by Digg users is