Affordable Dual Core from AMD: Athlon 64 X2 3800+by Anand Lal Shimpi on August 1, 2005 9:36 AM EST
- Posted in
AMD's Efficiency Advantage?Before we get to the actual barrage of performance tests, there is one issue that we have been wanting to tackle for quite some time now.
AMD has often argued that their dual core architecture is inherently more efficient than Intel's, primarily because of their System Request Queue (SRQ). All core-to-core transfers occur via this queue instead of over a main, shared FSB, which is the case in the Pentium D.
Johan put AMD's architecture to the test by measuring the latency of cache-to-cache transfers in AMD's dual core chips vs. Intel's. The results were quite impressively in favor of AMD's architecture. Cache-to-cache transfers on Intel's dual core CPUs took over twice as long as on AMD's dual core CPUs, but at that time, we could not find any real world benefit to the architecture.
Armed with a bit more time, we went through all of our benchmarks and specifically focused on those that received the most performance gain from dual core architectures. Using these multithreaded and/or multitasking benchmarks, we looked at the performance improvements that the dual core processors offered over their single core counterparts. For AMD, making this comparison was easy; we took the Athlon 64 X2 3800+ and compared it to its single core equivalent, the Athlon 64 3200+. For Intel, the comparison is a bit more complicated. The inclusion of Hyper Threading makes the single-core to dual-core jump a little less impressive in some cases, thanks to the fact that virtually all single-core Pentium 4 processors these days can execute two threads simultaneously. Thus, for Intel, we had to look at HT enabled, dual core and dual core with HT enabled, all compared to single core performance to get a complete picture of Intel's multithreaded performance scaling.
Remember that all performance increases are with reference to a single core processor, and in the case of Intel, we are talking about a single core Pentium 4 with HT disabled. More specifically, we used a Pentium D 830 (3.0GHz) for the dual core tests and compared it to its single core counterpart - the Pentium 4 530 (3.0GHz).
First, we have our Winstone 2004 benchmark suite; we omitted Business Winstone 2004, since it shows virtually no performance boost from dual core CPUs and instead, focused on Multimedia Content Creation Winstone 2004 and the Multitasking Winstone tests.
While AMD scales slightly worse than Intel (comparing the AMD Dual Core to the Intel Dual Core rows) in the MMCC Winstone test and significantly worse in the Multitasking 1 test, AMD scales better in the last two tests. Particularly in the third multitasking test, AMD gets a whopping 68.4% from the move to dual core while Intel only improves by 39.1%.
It is also worth noting that although Hyper Threading improves performance with a single core, enabling HT on the dual core CPU actually yields lower overall performance than if we had left it off (+24.1% vs. +39.1%). Johan explained exactly why situations like this exist on the Pentium D in his "Quest for More Processing Power".
Next up is the SYSMark 2004 suite. In all but two of the tests, AMD scales slightly better than Intel when going to dual core. The scaling advantages aren't huge, but they are tangible in some of the tests.
Once again, while Hyper Threading itself tends to impress, HT + dual core gives us a mixed bag of results, sometimes outperforming dual core alone while falling behind other times.
Finally, we have our application-specific benchmarks; here, we have AMD scaling better than Intel in 3 out of the 5 tests, but then in the remaining 2, Intel scales better.
Out of the 15 tests, 10 of them showed that AMD scaled better from single to dual core than Intel, while the remaining 5 showed the opposite, that Intel scales better. Out of the 10 tests where AMD offered better scaling, only 6 of them showed AMD outscaling Intel by more than a 3% margin (one test had AMD with a 2.9% advantage, but it was close enough, so we counted it). Of the 5 tests where Intel scaled better, 4 of them had Intel at an advantage by more than 3%.
While the Athlon 64 X2 does have much better cache-to-cache transfer latencies than the Pentium D, it appears as if for the most part, those advantages don't surface in real-world desktop usage. That being the case, the Athlon 64 X2 3800+ must outperform the Pentium D 830 based on the performance advantages of its individual cores in order to win this battle, not based on any dual core architectural efficiencies. So, does it?
Post Your CommentPlease log in or sign up to comment.
View All Comments
broberts - Wednesday, August 3, 2005 - linkISTM that the choice of Intel mobo is mistated. Further, IMO using an ASUS high-end mobo as one test platform and any Intel mobo as the other gives the former at least a 3% - 5% performance advantage. I wonder too, were all of the ASUS mobo's oc features turned off?
Our hardware configurations are similar to what we've used in previous comparisons. For this test, we focused on CPUs at or around the Athlon 64 X2 3800+'s $354 price point.
AMD Athlon 64 Configuration
Socket-939 Athlon 64 CPUs
2 x 512MB OCZ PC3200 EL Dual Channel DIMMs 2-2-2-7
ASUS A8N-SLI Deluxe
ATI Radeon X850 XT PCI Express
Intel Pentium 4 Configuration
LGA-775 Intel Pentium 4 and Pentium D CPUs
2 x 512MB Crucial DDR-II 533 Dual Channel DIMMs 3-3-3-12
Intel 925XE Motherboard [<======= Really???]
ATI Radeon X850 XT PCI Express"
robh3 - Tuesday, August 2, 2005 - linkRendering in 3dstudio max benefits from dualcore cpus. It however, also benefits from hyperthreading if this is a mental ray render. I'd like to see results for the pentium-d and amd-x2 when rendering in 3dstudio 7 with mental ray. I think the advantage is significant, especially in video renders as it's mathematically more effective to have more nodes rendering.
For your future reference.
MetroRider - Wednesday, August 3, 2005 - linkvery nice article indeed! know i just received three computers from Dell today with Pentium D processors, so i see the new X2 is a bit faster, but excited to try it out in general multi-tasking environments regardless.
One question i haven't seen answered is: How does a dual-core cpu compare against a true SMP dual-cpu system? Example would be a Pentium D 840 (or AMD X2 4200+) vs. 2x Pentium 4 - 2.8 GHz Xeon system? For the price, dual-core seems great. How does it handle running Windows 2000 Server or 2003 Server or running as an Exchange server?
If anyone has any input or has tried this, I'd be interested in knowing.
thanks again (sorry if the request is too off-topic)
fitten - Thursday, August 4, 2005 - linkOh... also, since AMDs cache coherency implementation and the way the dual cores work is more efficient than Intel's, so applications that have to pass ownership of cached data (L1/L2) back and forth a lot will be faster on AMD X2 machines. Of course, this type of application behavior is usually considered to be the result of poor design, so use well written apps :)
masher - Thursday, August 4, 2005 - linkActually, I'd give a slight (very slight) edge to Intel's approach on SMP systems....though AMD's Hypertransport is certainly far cleaner on paper, it hasn't paid off quite so well in practice. However, HT is considerably more suited for a NUMA-based multi-cpu architecture than for SMP. Unfortunately, there aren't many NUMA OS's out there at the moment...
fitten - Thursday, August 4, 2005 - linkOne question i haven't seen answered is: How does a dual-core cpu compare against a true SMP dual-cpu system?
Depends. Dual CPU systems and dual core systems both have two cores running in them. The main difference is that two single core AMD machines (two Opterons) can have independent memory banks if you buy an appropriate motherboard. The dual core systems 9(both Intel's and AMD's) share the same pipe to memory. This isn't as big of an issue for AMD parts because the Athlon64 (single core) really doesn't make good use of dual channel memory anyway (dual channel over single channel is at best 20% faster in memory intensive apps while most see around 5%) so neither of the cores in the dual core setup get memory starved unless both cores are running full tilt (by benchmarks, I guess this would mean that if single channel were 1, dual channel were 2, the two cores running full tilt would need 2.4 channel memory to never be starved). The Intel parts, however, will suffer more. Dual channel memory on a P4 can be saturated by a single core so dual cores going all out can become starved a bit worse (one core would be 2, two cores would need 4/quad channel memory to never be starved).
So, your performance will depend on what types of applications you will be running.
Tujan - Tuesday, August 2, 2005 - linkWell Im missing the reasons for comparing the 640 Intel,and 630 D Intel when both are less than the AMD 3800+,and 3800+x2.
True if you where to take the 'best of'in performance to price between both Intel,and Amd the 640 Pentium 3.2 would be the comparitor .
Performance alone though,true AMD 3800+ surpasses the P4 640 .The 3500+ AMD 64 would have been a closer comparitive if solely on a price basis.
With both the AMD 3500+,or the Intel 640 to be either choice in that price point to have. They compare in performance as well.Especially since the next Pentium pricepoint the 3.4 Ghz 650 is the closer comparitor.
The 3800+ AMD 939 single is still more expensive than the Dual-Core 830 D . With the newest AMD 3800+ still at the 400$ spot.
Comparing the Intel 640 to the 3800+AMD is lopsided.
The Intel 650 is the closer comparator,in terms of cost.Still though the Intel 640 remains the gulf between them when cost is the comparator.
Im missing reason author chose to compare these specific CPUs. Since the Intel is clearly not in same performance class.Although for even reach of benchmarks,would be the best'of''in price points to beat.Should have compared the Intel 650,to the AMD 3800+.
AMD 64 3500,AMD 64 3700 would have been closer comparitors to cost between the two,to match up performance.Seen the benchmarks before between them.AMD winning some,Intel winning some.Howebeit,the Intel has HT.etc.
AMD 64 3800+ CPU is a tough machine.AMD 64 3800 X2 twice as tough.
Anybody know what the transistor count is for the Pentium D 830 ?
coldpower27 - Tuesday, August 2, 2005 - linkThe Smithfiled is 230 million transistors, it's interesting though, AMD more then doubles transistors count going from Single core to Dual core, while Intel is the opposite.
Tujan - Tuesday, August 2, 2005 - linkThanks for reply. Can you give me the link/URL for a chart on the newer CPUs with transistor counts included. Or can you tell me for example,the website,to go to to look at the chart ?
These new processors are simply nummy.
redhatlinux - Tuesday, August 2, 2005 - linkOnce again, I believe Anand has shown un-biased data. I run a Dell hyperthreading rig, just because its convenient. BUT, as others have pointed, the X2's were designed from the ground up to be dual, on chip. To have to leave the core, and use the FSB, just shows that Intel is playing catch up. Benchmarks can, and often do show whatever the person running the benchmark wishes to show. Anand has ALWAYS, IMHO, tried to show, unbiased, accurate DATA. The basic design of the long piped Pentium will almost always favor Intel in benchmarks which, in essence are SIMD, in nature. That is, processing streams of data, with a single instuction. Price points, for ANY given CPU, GPU, HDD etc will always be Market Driven. The complexities, which go into the pricing decisions are well beyond the scope of this forum IMHO. AMD will continue to offer innovative design, over Intels 'brute force', more megahertz must be better, approach.