PC Performance Tested

19

Comments

+ Add a Comment
avatar

music2myear

Thanks for the comparison. It does appear to be a rehash of older information, but it was nice to find this at the top of the page as I'm building a new box and needed this precise information.

+1 for relevance.

However, there are a few issues:

1. The picture of the CPU you have labeled the i5 very clearly has i7 stamped on the die.

2. In the results paragraph you say "For first place we're proclaiming the Core i5-4570K the midrange king…", which, if I'm not mistaken, isn't one of the CPUs in this comparison. The i5-4670K, on the other hand, might be what you meant.

avatar

UtilityMax

The AM3+ socket seems to be just as dead as LGA1155. AMD keeps quiet about future FX CPUs, but the recently leaked roadmaps show nothing new for a year or two, which probably means end-of-life.

On the issue of Intel vs AMD, I think the true hero in the value segment right now is AMD's FX 8320, which easily overclocks to the speed of 8350, but costs less than even the entry-level Core i5s. All in all, AMD's Vishera line is still surprisingly competitive for the money, considering it's based on almost two year old architecture. I think AMD has the capability to battle with the Haswell Core i5 if it only made a Steamroller-based 8-core FX CPU, but for some reason it doesn't

PS: Can the editors fix the CPU benchmark table? Some lines have more than one number in bold. Some lines just don't make sense, like the Handbrake test.

avatar

duken

The tests are nice, thank you. The conclusions might be skewed by the battery of tests however. If anything, the results show that the best CPU depends on what you plan to do with it.

For gaming and general floating-point computation, Intel wins! No argument about it. On the other hand, if you're doing integer computation as in the 7Zip test, the 8 integer computation units in the AMD easily beat the Intel CPUs. (Integer computation is used, for example, for hashing, encryption, compression, fixed-point signal processing.)

At the core of the whole thing is the shallow understanding of the CPU architecture. The FX-8350 is not a 8-core CPU (I know that AMD market's it as such, but that's wrong), rather it is functionally equivalent to a quad-core with hyper-threading. BTW, hyper-threading is Intel's term to mean that certain functionality (e.g., instruction decoding, register sets) are duplicate within in core, which allow some parts of the core to keep working while the other parts are waiting for something to happen. The Intel CPUs duplicate very little, but in the FX-8350 the only part that's shared is the floating-point arithemetic unit and, since historically that also AMD's slowest part, that's why that CPU falls behind in tests heavy in floating number calculations.

I not arguing for one CPU versus the other, but I hope that MaximumPC guys get to figure out a little more of the details so that they can tell the reader the whole story. If the set of all tests is a accurate representation of the general user, then I agree with your conclusion. Otherwise, *for the applications I run*, your tests clearly tell me to buy the FX-8350.

avatar

vrmlbasic

How is AMD's module system similar to hyperthreading? Hyperthreading can only run one thread at a time, processing another thread only when the primary thread is waiting for something, no? While Piledriver & Bulldozer cores have the shared decoder they can still run 2 threads much more "simultaneously" than can hyperthreading.

As I recall, the FPU can be halved and each core can use half, meaning that it is only shared when it is advantageous/necessary to use the full FPU for a task...or if the software is too stupid to understand AMD's system (which seems to be the case even now, years after Bulldozer hit the market).

avatar

duken

Did you carefully read my full comment? If anything, I praise AMD for doing hyperthreading right (call it dual-threading or whatever else you like), and I pointed that I thought that the tests overly skewed towards FPU-heavy applications. For gamming, that might be appropriate, but that is still a skewed picture of all computation heavy uses for a PC nonetheless.

Personally, I'm actually biased towards AMD. But I'm also realistic. Unfortunately, at the moment, Intel CPUs are better for gamming because games use alot of FPU computation and AMD is lagging behind in that part. I hope that AMD can roll up their sleeves and give Intel a hard time because we need competition to ensure continued investment in innovation.

avatar

vrmlbasic

I did and I believe it to be incorrect: an AMD module is not "functionally equivalent" to a single Intel core with HyperThreading enabled, especially not if we're including Steamroller.

avatar

tenaomalo

@Ninjawithagun, I'm not sure what article you're reading but they didn't compare an APU with any Intel Processors. They compared an FX-8350 to them. In trying to provide accurate information, yes, overclocking on a APU does come up short compared to much higher end i5's, since APU's are geared toward lower-end, entry level gaming systems. I have an FX-8320, which doesn't even include integrated graphics (Which is needed to be an APU, by defenition), and have that overclocked to 4.5GHz stable, with no increase in voltage whatsoever, running at around 40 degrees or so on a CLC. So, while I agree with your statement about APU's not being able to overclock better than an i5, please re-read the article :)

avatar

dashize

this article is almost 6 months old. It was July when they did the test and it was in the October Mag. Is there really nothing new to post?

avatar

Ninjawithagun

MAJOR FLAW. The GTX780 costs $499...NOT $650 ;) For $650, you can buy a GTX780Ti, which would have outperformed the two GTX660Ti cards in most benchies.

TYPO: "So where does this leave us? For first place, we’re proclaiming the Core i5-4570K the midrange king by a margin wider than Louie Anderson." Should read, "Core i5-4670K", not "Core i5-4570K".

Also, you failed to capture the true essence of purchasing an Intel CPU...overclocking ability. The AMD APUs do not have as high a ceiling for OC performance benefits due to TDP limitations. Out of the box using mild OTS CPU liquid cooling, I am easily able to achieve a 4.6Ghz overclock on my Intel 4770K. With a bit of tweaking, I am able to get a solid 5Ghz overclock with manageable temps under full load benching using Prime95. No way can you do that with an AMD APU without a serious watercooling system or extreme phase-change cooling.

And let's not forget that clock-per-clock cycle, Intel CPUs have proven more powerful than AMD APUs. Even pricing is not competitive anymore with Intel lowering the prices on all of their Ivy Bridge and Haswell CPUs recently. Broadwell will be out in less than a year, so I hope AMD is ready for another beating ;)

Bottom line, pay attention to detail and UPDATE your article before reposting it online 6 months later. Laziness is for those who want to be unemployed!

avatar

JimmyD1964

My 8350 gets 4.7ghz easily on AIR. So you get 4.6ghz with the Intel on Water? Sort of comfirms the heat issues the Haswells are having with 22nm huh?
And At $334 vs my $189. I can almost get 2. Or go ahead and buy all my ram with the difference.
Confirms AMD's bang for the buck. Enough said.

avatar

vrmlbasic

The AMD chip in the CPU section here is not an APU.

avatar

AFDozerman

Jesus. Vishera launched in 2012? It's 2014. The hell AMD?

avatar

vrmlbasic

Nice article. The IDE vs AHCI was jaw-dropping.

I'm definitely interested in more SSD RAID vs Single-Drive stuff as SSD RAID is territory that doesn't seem to have been fully mapped.

The PCIe 3.0 vs 2.0 was unsettling as AM3+ is still only on PCIe 2.0 and since it did appear to be holding cards back I'm concerned about the R9 290's crossfire over PCIe 2 :(

avatar

JimmyD1964

PCI2 vs PCI3 = about 2 to 3%, Google it.

avatar

Slugbait

I found out the hard way about IDE vs AHCI seven years ago. I got an 80G Raptor during a wootoff for just $95 shipped (newegg had the next best price: a 74G for $155 shipped).

I didn't know about AHCI, and was a little pissed that my system was performing no faster than when using my larger/cheaper 7200 hard drive. After a little googling, I found out about AHCI and re-installed...blazing speed.

Anytime I see forum posts from people complaining about the performance of a new hard drive, AHCI is always the first question I ask. The vast majority of the time, AHCI is the solution.

I retired that Raptor (and most of that machine) to my HTPC four years ago. It still tickles me how fast that machine goes into and out of S3 sleep.

avatar

maleficarus™

Great article! Well done!!

avatar

EKRboi

These are the types of articles that keep me coming back to MPC. Thanks guys and gals!

avatar

vrmlbasic

Hey MPC,

For the CPU benchmarks you bolded the Intel chip, indicating that it won, on the handbrake blu-ray encoding when it looks as though the AMD chip actually won.

9042 (intel) vs 8400 (AMD) seconds to encode

avatar

jimmthang

Ah, good catch! Fixed, and thanks!