The 8 core Intel thread.

[gpuerrilla]

Its total raw performance only matches an overclocked 3770K, which cost 3 times less (Not to mention the board price)

Why would you choose the same performance at triple the price?
If I was looking at it from an enterprise POV I'd buy two of the 4c/8t Haswell Xeon's at 200 a pop and build two systems at the same price and about 66% higher performance output.

Or, when Haswell-E comes, you'd just pick one of those Octo Cores while probably coming in at nigh on the same price.

Yes, there are things that can benefit from lots of cores and threads, but unless you get a stellar deal second hand, you'd be far better off with a 4930K than one of these Xeons, and have 300 quid spare.

The end performance of this Xeon at stock is that of a 4.6GHZ 3770K, that's performance available at 230 quid, whereas this CPU generally costs 3 times that.

I imagine you'd take two of these Xeon's and chuck them in ; Asus Z9PE-D8 WS Dual Socket C602

But then you're basically spending a crap ton of money to get the same end performance you're going to get from a single overclocked 8 core Haswell-E.

If anyone remembers on socket 1366, you could get some overclocking Xeon's, that was basically i7 920's? I remember someone basically selling 2 of those xeon's with the board for about 200-300 quid.

The way I see this is pretty hypocritical. For months some of us argued on the low tier CPU ground that the FX was slightly cheaper and offered similar performance (or some cases slightly worse) than it's i5 equivalent.

Your point stressing that as it was intel, you would pay the extra. This is one of the reasons martin why I find you as someone that likes to argue and act clever rather than dismounting off ones high horse and taking it like a man. As the saying goes: If the cap fits...

 

[JamesM]

That's just the way IBT calculates it, I'm presuming it divides the overall result by the number of threads so non-HT processors tend to get a bigger score even though they'll be doing less.

Has OP come to his conclusion that Haswell-E 5xxx will be overkill and everybody should buy AMD yet?

Not sure if serious......

He has 2 Intel rigs and 2 AMD rigs. Perhaps instead of being 'fan boi' he sees plus points with both sides?

Not sure where all this "CPU A will blast CPU B" stuff came from. Tell me if i'm wrong but the whole point of Andy is doing this is purely to find out just how well software and games makes full use of the 8 Intel cores to see whether the newer Haswell-E is worth it. He already has an 8 core AMD setup so wanted something to compare against.

 

[ALXAndy]

Right, I'm back from holiday 1...

OK, CPU arrived, board arrived. Had to run up the high street like no one should to get to the PO in time to get the CPU before they closed for the day (Saturday) but got it.

Board and CPU work fine, even on the old bios. Have updated it now. Have spent the past two days rebuilding the rig and gathering data.

The upshot is this.

I can't overclock the BCLK *at all* not even 101mhz. Pretty disappointed about that, was hoping to get at least 150mhz over stock.

The Intel beats the AMD at 4.7ghz by about 5-10%. Up the AMD's clock to 4.9ghz however and it turns the tables by 5-10%. However, in order to beat the Intel it *must* be ran at 4.9ghz and with the weather the way it was lately I had to literally take the PC apart and aim a floor standing fan into it to stop it from shutting down.

There are things the AMD wins at at either clock speed, and there are one or two things the Intel wins at no matter what the clock speed.

Data to come soon, though I will be concentrating on the 4.9ghz results as otherwise it was just too much data to sort out. I am now moved on to core use and threading, with some very interesting results thus far.

I'm slowly beginning to build a mental picture of what Haswell E will be like, though. And like I said in the other thread it's all going to come down to how much support Intel can muster up for it. Without it? you'll be better off with a I5.

 

[mmj_uk]

He has 2 Intel rigs and 2 AMD rigs. Perhaps instead of being 'fan boi' he sees plus points with both sides?

Try looking at his post history. I rest my case.

You can't blame him for buying some Intel processors they are just better as it stands but being an AMD fanboy he won't admit to it. He's top poster in the Haswell-E thread but all he has really done is try to put people off buying it.

 

[Martini1991]

The way I see this is pretty hypocritical. For months some of us argued on the low tier CPU ground that the FX was slightly cheaper and offered similar performance (or some cases slightly worse) than it's i5 equivalent.

Your point stressing that as it was intel, you would pay the extra. This is one of the reasons martin why I find you as someone that likes to argue and act clever rather than dismounting off ones high horse and taking it like a man. As the saying goes: If the cap fits...

The Xeon in the OP is over 700 quid, the i7 which offers better performance being 240.

That's completely different to arguing the difference between the FX8350 and i5 4670K (Which until recently, the difference was about 30 quid, it's now closer to 50 quid).
If you go to another thread, I said I'd pick the FX8320 over the Xeon that Andy's picked.

So not hypocritical at all.
I do like to argue, I don't need to act clever, I know I'm clever, which is why people like you come up trumps to me in any logical debate (And if I wasn't a logical thinker, the things I speculate on wouldn't turn into fact as regularly as they do)

I only saw your message because I'm at work, so I'm not logged in, felt I'd "clear it up".
Feel free to twist some more stuff against me.

 

[ALXAndy]

Try looking at his post history. I rest my case.

You can't blame him for buying some Intel processors they are just better as it stands but being an AMD fanboy he won't admit to it. He's top poster in the Haswell-E thread but all he has really done is try to put people off buying it.

Have a good look through my posts. They're usually in response to trolling blanket posters like yourself.

In your eyes there is only one product. Only one manufacturer. The other one? haha ! that's just a way to sit and score points all day.

If you actually read what I posted instead of shrugging it off you may learn something.

"They are just better". That's exactly the sort of BS I am referring to. It's funny, because on other parts of the internet and other forums people are actually intelligent enough to digest information and make their own well researched decisions but here? jesus, it contains more nonsense spouting trolls than any other forum I post on.

I just call things as they are. Haswell E = Haswell with more cores, running a CPU that Intel have been selling for years only £900 finally gets you a CPU that has 8 cores *and* you can overclock. And Intel should be lauded for that... Why?

If they did great things then they would get the praise they rightly deserved. But they don't, and they haven't since the Clarkdales. Ever since then they've gone around with their padlock railroading people into stupidly expensive products that yeah, are faster than AMD but define better? oh yeah a CPU that costs over double and yet doesn't even go near offering double the performance.

See, troll, when I put a CPU in a rig and begin testing it I do it open eyes and open mind. I run the tests, watch what happens. So I just saw an AMD beat a chip Intel were charging £1000 for. Oh no, the Intel lost. Why? because the Intel is locked.

Don't tell me ! that's the reason huh? the Intel is locked and so can't put up a fair fight for itself.

Whose fault is it?

That would be Intel's.

Do you see?

But no don't worry, they're about to make it all OK by releasing a £900 CPU.

And I'm supposed to be peeing my pants, why?

 

[Martini1991]

Unsurprisingly I was right about a 4770K at 4.4GHZ beating the Ivy 8 Core 2GHZ (Although it was simple logic, and anyone could have called it), I figured I'd take a peak at your results you've just posted Andy.

 

[ALXAndy]

Unsurprisingly I was right about a 4770K at 4.4GHZ beating the Ivy 8 Core 2GHZ (Although it was simple logic), I figured I'd take a peak at your results you've just posted Andy.

Yeah no surprises tbh dude. However, have a look at this...



13.9 watts at idle, even though I have disabled speedstep and forced on the 4 bin turbo to make it run at a constant 2ghz.

It's also pretty much dead neck and neck with the AMD @ 4.9ghz, and that consumes about three times the power. Which means I've been able to drop all of my intake and rad fans to 5v meaning the rig hasn't really changed, but now makes pretty much no noise at all.

Will sort out the results and info soon. Gotta say though, you'll be amazed at how many games use all of the 16 threads.

 

[JamesM]

I presume BF4 will be one?

Do the extra threads make up for the lack of IPC compared to say the 4790k?

I saw in the CineBench thread that it only scored 710 which is lower than my overclocked 4690k at 4.7GHz. Not much lower mind

 

[Martini1991]

Haswell made some gains in FP performance, so Cinebench sees some much higher gains that it would over Ivy usually.

Strange to refer to an Ivy as lack of IPC, still the second highest IPC in the world.

What use is it if a game can use 16 threads but end up coming up trumps to a 4690K? (Which I assume the Xeon will in most except extreme cases as far as gaming goes).

And which of those games that use 16 threads are merely load balancing (Since you've got 8 slow cores)?

 

[ALXAndy]

Haswell made some gains in FP performance, so Cinebench sees some much higher gains that it would over Ivy usually.

Strange to refer to an Ivy as lack of IPC, still the second highest IPC in the world.

What use is it if a game can use 16 threads but end up coming up trumps to a 4690K? (Which I assume the Xeon will in most except extreme cases as far as gaming goes).

And which of those games that use 16 threads are merely load balancing (Since you've got 8 slow cores)?

I recall running the Metro Last Light benchmark oo, about a year ago. A mate of mine was running a 3770k, overclocked, the AMD put out a higher FPS count.

I ran it on the AMD before removing it, then ran it on the Intel. The Intel won. Not by a massive margin, but it won. The game saturates the entire 16 threads..

I guess that it basically comes down to making yourself a match. There's an age old article on Eurogamer that basically has the devs come out and say that you're better off with an 8 core CPU, ignoring the clock speed (because it will slaughter the console counterpart) and just doing your best to match up the spec of your PC to the spec of the consoles (cores and a shed load of VRAM).

TBH Martini I really don't know what CPU is beating what here. At the end of the day what matters (like, what really matters) is that a CPU can put up a show for itself and run a game smoothly at an acceptable FPS rate. Once it does that? then anything else is pretty much splitting hairs.

I recently switched out a 4.4ghz AMD for a 1.78ghz hex core Westmere. This time I did not bother running untold benchmarks, I just plopped the rig down in front of the telly and invited the guests to play what games they wanted. It puts up a marvelous show, especially considering it's running SLI 670s. But the point is that basically this £48 CPU is more than good enough for gaming, especially when it's threaded properly.

Whether or not that makes it worth spending £900 on an overclockable Intel? tbh, no, it's not. That's probably why Intel concentrate more on Z97 than they did on X79, for example.

I'm already starting the basis of some predictions on what sort of benchmark results Haswell E will put out. Mind blowing for sure, but whether or not the price matches the performance? can't see it. Intel aren't exactly known for giving away Easter Eggs.

/Ray Liotta, Goodfellas..

"You want cores? **** you, pay me".

 

[Resa69]

Haswell made some gains in FP performance, so Cinebench sees some much higher gains that it would over Ivy usually.

Strange to refer to an Ivy as lack of IPC, still the second highest IPC in the world.

What use is it if a game can use 16 threads but end up coming up trumps to a 4690K? (Which I assume the Xeon will in most except extreme cases as far as gaming goes).

And which of those games that use 16 threads are merely load balancing (Since you've got 8 slow cores)?

Yes these questions need to be answered!^

 

[Martini1991]

Were you and your friend using different GPU's for Last Light?

I'll be honest, anything you say I'm skeptical taking it at face value.

First result for Metro Last Light Benchmarks brings up ;

http://www.techspot.com/review/670-metro-last-light-performance/page6.html

And if you look at overclocked results, the FX83 comes up trumps to the 3770K.

The game uses threads as covered by the text, but the 8 threads on the FX83 come up trumps to the 4 of the i5.

I'm not ignoring the part where you said your Xeon one upped the FX83. It just wasn't applicable to my question.

Looking at MLL, the engine seems capable of utilizing threads (I'm surprised it can do 16 that said). If it really does max out the CPU, 100%, then if we do some math, The 4GHZ 3770K being essentially what your CPU can attain in raw performance, then it would best a clocked FX8350 (In which case, the 3770K would still trump the FX83).

Unless, your FX83 was overclocked, and the i7 3770K wasn't, but by Techspot's results, even at 4.9GHZ on the FX83, assuming perfect scaling, it'd be more parity/if not losing than besting.

 

[ALXAndy]

We were both running GTX 670 SLI at the time. The AMD was at 4.8ghz, his 3770k was at 4.6 (which is his limit, got a duff chip).

Looking at the chart you posted stock there is about 5 FPS in it. But overclocked.. Things change. At stock the 8320 loses around 40% of its Cinebench R15 score..

I'm still compiling data on core use. Two more games to go should about do it. At least then it'll settle the "But hardly any games can use more than 4 cores" argument.

Ed. I guess what it came down to was overclocking headroom. The AMD has far more of it from the stock speed to the max overclock than the Intels do. Maybe that was why.

 

[Martini1991]

Your maths is failing you.
4GHZ to 4.8GHZ is less overclocking than 3.5GHZ to 4.6GHZ.

In total overclock and percentage, how is that in anyway more headroom to the FX83?

Also, when overclocked each to 4.5GHZ, their figures are 13 FPS difference to the advantage of the 3770K.

You could post 10 games using more than 4 cores, in the grand scheme of things, that's still "hardly any games". Bearing in mind in total I have around 500 myself (432 on Steam, some on Origin, and then discs). Even 50 games is 10% of my library, that's still "hardly any games use more than 4 cores".

 

[ALXAndy]

Your maths is failing you.
4GHZ to 4.8GHZ is less overclocking than 3.5GHZ to 4.6GHZ.

In total overclock and percentage, how is that in anyway more headroom to the FX83?

Also, when overclocked each to 4.5GHZ, their figures are 13 FPS difference to the advantage of the 3770K.

You could post 10 games using more than 4 cores, in the grand scheme of things, that's still "hardly any games". Bearing in mind in total I have around 500 myself (432 on Steam, some on Origin, and then discs). Even 50 games is 10% of my library, that's still "hardly any games use more than 4 cores".

Your knowledge fails you. The 8320 is a 3.7ghz CPU. It only boosts one core to 4ghz, so that happens pretty much never.

It's the same on my Xeon. Says it has a max multi of 23, so 2.3ghz, then you see this..



Hence, the AMD can do 1.2ghz fully bench stable over its original stock clock. Is there 1.2ghz left in the tank on 3770k? I doubt it.

That's what I was getting at dude.

Ed. Sorry, still crunching loads of data :S

Going back to the core argument. I can post you oo, at least ten games that are using them. And yeah, ten games is still a small amount, but the important part to garner is that they are the last ten games launched for the PC with any clout. All of the big titles, all of the big franchises. That says a lot for the future. It's not like they're going to change their mind and relinquish core support is it?

 

[Martini1991]

I thought you had an FX8350. That's more of an assumption fail than knowledge fail, the knowledge fail is yours

The FX8320 isn't a 3.7GHZ CPU, it's 3.5GHZ.

In which case the overclocking between the 3770K and FX8320 is more of a muchness (Although I'd wager on average the FX8320 does 200MHZ more) as 4.6GHZ on the 3770K isn't "dud" it's average.

Either way, it doesn't explain your disparity at all.

And pretty much everyone has agreed going forward games will start using more cores. But showing me 10 games that use more 4 cores does not in anyway change the fact that "hardly any games use more than 4 cores" So, it doesn't put it to rest at all.

 

[ALXAndy]

It probably doesn't explain the disparity no. PCs are strange beasts at the best of times, and results vary tremendously from one reviewer to another. Tiny **** logan usually ends up with results that match up to no one else. Maybe because he's trapped in the past using Windows 7?

I find it funny that people are running quad + HT CPUs and insisting on an OS that doesn't even handle their CPU properly.

I also know that apparently SLI hilariously works better on AMD than Crossfire does. Was reading about it on Tom's..

 

[Martini1991]

TTL's results come under scrutiny constantly, it matters not what his results say if no one trusts them.

 

[Svyper1]

so basically intel beats AMD, no surpise...given that he CPU you chose is not very extreme and designed for servers running low spec but many many things at once...

Whilst the cpu costs more you can pretty much gaurrantee of low power cost and long life span and reliability and low low temps in tight spaces. pretty much what a server needs...

AMD at 4.9 is an oven whats the point...you can't run a server on that 7/7 365/365