I7 920 vs Phenom II 965 with an ATI 5870.(Finally!) - Page 4

Last message on previous page:
@ Jimmy how do you know that the AMD system didn't have software errors? How do you know that it didn't have AA turned on in CCC, or the Catalyst AI set to max, or frankly a whole list of possible issues that a *bad* reviewer simply wouldn't realise.

Are you saying you believe these benchmarks more than anand's?

No but considering that I have a HD4870 1GB I can tell you that unless they went in and turned it on, the AA is set to Application settings BY DEFAULT. Its how I run my CCC, which I love. CCC is the most innovative GPU control system ever. Even though the Overdrive isn't as good as others it is nice to have built in with fan control (mines at 71% always still pretty quiet).

The benchmarks are interesting but unless you can see it you have to take them with a grain of salt, even the ones you trust. But you don't seem to trust any of them except if they can prove your point.

My point is simply this: The Phenom II had been out a while. Asus is known for quality mobos and it had even the latest SB750. So unless the Windows install magically only went bad on the Phenom systems there is nothing wrong with it.

Now if they do the same tests I am sure that performance would go up but that still wont change the fact that a Core i7 is able to push a GPU pretty damn far. And a 4890 has a 200MHz advantage average GPU and about 100-200MHz memory.

I trust Anand even though a lot of AMD fans don't, you would be the first who does really. Most AMD fans think Anand is a bunch of paid Intel schills like they think THG is.

Still the THG one might have been biased but still the Core i7 was at a disadvantage.

Meh I wont argue anymore. You are right, I am wrong. AMD For Life and so on and so forth. I shall never buy Intel again even if they show the best performance in everything (considering I do rip DVDs and edit videos for games but meh).

Oh Jenny,

Almost all of your posts in this thread were of you trying to figure out WHY the i7 would end up slower than a Phenom II when a GPU bottleneck was present. Now that I relay all the evidence to you.. all of a sudden you no longer care why (because caring why would be you apologizing to me for being rude to me when I was clearly only attempting to spead the evidence).

The results you claim are garbage where the results YOU YOURSELF posted. Once I used YOUR source to refute your arguments.. now they're all of a sudden "Garbage results". Oh the hilarity ensues.

You have used no facts. You have simply used assumptions (your assumptions) rather than consult the evidence. I am the one who has used facts (evidence) and even supplied links to my evidence for your careful consideration (you have supplied nothing but your own assumptions based on your own observations of a graph a.k.a your opinion).

As for a sensible gaming CPU, the difference between a Phenom II and a Core i7 in gaming scenarios is minimal. I would never recommend a Core i7 machine to someone who simply wants to game (doing so would be tantamount to being dishonest). But when you factor-in all of the other tasks we use computers for on a daily basis the picture might change and the recommendation made will need to factor this additional information.

And this is why I made my statement that for CPU Intensive tasks.. a Core i7 is superior in every way over a Phenom II. This statement is indeed factually accurate (as the current issues when gaming stem from a Graphics card bottleneck and a communications conversion process between the PCI Express bus and the QPi link which does not count as being a CPU intensive task).

Peace.

"Benchmarks where there are now graphics bottlenecks point to a fact that i7 is faster. Your statement is only true in the context of a GPU bottleneck."

DUDE NOBODY CARES! Nobody cares if he gets 10 measly frames more if he plays on low DAMN IT!!!

Would you call 1680x1050 4xAA low? Or 2560x1600 no AA / 4xAA? At 2560x1600 4xAA all the CPUs are limited by the single hd4870x2 but when you add another card i7 pulls ahead.

http://www.tomshardware.co.uk/core [...] 673-7.html

I have been consistently talking about high resolution/high detail gaming.

I see. Well they do contradict your position so it is understandable. Looks like this isn't going anywhere but atleast with ElMoIsEviL's posts any consumer trying to actually make a decision has solid information to base their decisions on. Add to that my and others 'not so technical' contributions, I think the information is out there. All I would like to add is that the evidence really seems to converge to one conclusion especially when you factor in the application benchmarks.

So basically any interested consumer, that allready know what they are going to use their computer for and what their GPU budget is, should be able to decide which CPU is better for their needs.

So if evidence contradicts your position then the evidence, no matter how credible, is wrong?

What you're saying here is that your opinion and world view trumps facts.

Err.. ok.

What specifically is wrong with tom's benchmarks? Have you checked the evidence that you think supports your position? Specifically what kind of graphics hardware they were using compared to tom's?

All right, if you have nothing better to do:

http://www.modreactor.com/english/ [...] rhead.html

http://www.overclockersclub.com/re [...] resh/9.htm

http://www.neoseeker.com/Articles/ [...] 65/10.html

http://www.guru3d.com/article/phen [...] ew-test/18
Strange that these results tell a weird story. In RE5 the phenom II gets squashed, but in the other titles it's pretty close, or AMD wins at the higher resolutions. They do make note of the 5870(s) needing a more powerfull CPU to maximize its potential though in the conclusion.

Here they show the i5 750 beating the PH II 965 pretty easily when both were clocked at 4.0ghz. http://www.xbitlabs.com/articles/c [...] html#sect0

PS: Cyber romances are lame. (They know who I'm talking about)

http://www.neoseeker.com/Articles/ [...] 65/10.html

I7 gets crushed by a 965 BE in Bioshock and Farcry 2. Now I dunno about you, but that Bioshock result doesn't look gpu limited to me.

There are other factors involved here. It's not a simple case of clockspeed or anything like that, there are other factors involved that lead to i7 failures in a lot of games - we've seen it in plenty of ocassions in the past especially with nvidia cards.

http://www.tomshardware.co.uk/gefo [...] 510-2.html

The really interesting thing about that old review - which caused a major amount of intel fanboys to get up in arms - is that the Farcry 2 results with the 4870 x2 were seen as being the 'proper' results...even though it was a driver issue holding the phenom II back. Do those results today and you'd have the phenom II ahead on the nvidia hardware and equal on the 4870x2.

Well, I think in 4 more gpu generations, THENS when we'll see the i7 kick butt.
Maybe 5?
The new gen is just like the old gen, the numbers are higher in games on the particular gpus, but, arent that different from 1 gen to the next between the 2 cpus, using both gpu gens.
Whys that?
Oh, wait for the next gen?
As to the using a miniscule res to prove anything, then it becomes a pure benchmark, and not a real world app, doesnt it?
Nope, the i7 at best will only show its max IPC difference, which basically means squat for gaming, as a newer card or cards will give much better perf, or equivilent, especially with better cards, or this gen, not next.

Just to point a little thing out about that modreactor review -

http://www.modreactor.com/english/ [...] 7-920.html

There are the system setups, and they look to be as fair as can possibly be, down to the same speed memory used. The i7 gets flattened by the 955 BE, really quite badly.

I'm not saying I believe this is a perfect bench, but they sure as hell look more valid than those very strange benchmarks done by legionhardware.

You just don't seem to be paying attention with what people are telling you. That test is done with a single gtx260... I can't understand how you might even for a second think that those bioshock numbers aren not GPU limited. Considering that overclocking the CPU had no positive effect...

Probably the first thing I agree with you in this thread. In my opinion you're just trolling though...

That's the impression that I got though. Or that they were supposedly atleast equally powerfull.

And what about the other test with the two 4890's? Or are we ignoring that now too because it goes against everything you have been saying? Maybe I need to quote it again?

Is it getting through this time? When you see results like this it's no wonder intel gets accusations of buying reviews aimed at them.

What? That single datapoint somehow proves you right? You seem to completely ignore my posts by the way...

EDIT:
Are you sure though that they are not GPU limited? It is after all highest quality and pretty high resolution. And with single card they score 20 fps each where they are clearly GPU limited. Phenom II scales very well. Pretty much as well as one would expect. 65% increase. So I wouldn't even say this one datapoint goes in your favor. It has been established that sometimes i7 doesn't handle itself as well in GPU limited situations allready.

I wholeheartedly agree. What makes them even more untrustworthy is that many review sites tend to use built in benchmarks instead of actual gameplay when possible. Differences between i7 and pII seem to shrink when actual gameplay is tested. But all in all in my opinion i7 is a bit faster. That's where benchmarks as a whole seem to converge. As for future proofing? I don't think it makes any difference. They are still so close together that if one is unplayable in a game then likely the other one is also. I'm relatively certain that both of them will suffice untill next generation of CPUs in 2011.

I don't especially trust or distrust Intel. I know it's a company and it will look after its own interest. If evidence of this corruption comes to light then it is a different matter. As for older benchmarks looking bad and sharing wealth: I guess there might have been some hardware and/or software incompatibilities that were later solved. I remember seeing some very unflattering i7 benchmarks. They might have been paying atleast to people who code drivers to speed up their work. But I haven't seen any evidence that they started bribing reviewers. Besides that is a dangerous practice in a sense that there are pretty easy ways to replicate the results using same hardware and software. It isn't a prohibitive cost for say couple of friends even who are planning to upgrade their computers at the same time.

Not sure anymore if you're trolling or not. So: Apologies for the remark.

if phenom beats i7 in some games at high resolution doesn't mean that a phenom is faster than the i7, because GPU is casing bottleneck and is not able to keep up with the cpu, there for, cpu is out of the equation.

if you really wanna know the reason then lets break it into parts, Intel uses QPI to connect with the GPU memory controller inside north bridge at 4.8GT,while AMD uses HT instead, now during that operation there is a little favor going to AMD mobo than Intels which gives AMD system 1 or 2 more fps, now that doesn't mean that phenom CPU is faster, but the way their motherboard and GPU connects is a little different.

However if you still believe that phenom is faster than i7 at stock speed (or even overclocked phenom 965 losses to i7 at stock speed), then you should take a look at this where we have i7 alone vs phenom alone in applications.

http://www.tomshardware.com/review [...] 389-6.html

Here is a link. Now, before going to this link, keep in mind there are equally clocked CPUs with minimal variables. Clock for clock Yorkfield is faster than Deneb.

****Go to the end of the review which shows a Q9770 vs i7 965.**** Q9770 > PII 965

http://www.guru3d.com/article/core [...] e-review/7

i7 is a clear choice when faced with a CPU limitation when one wants maximum performance, even if FPS are over 60.

PS. I also do a lot more than just gaming.

Unless its on a graph, the low end fps means nothing, as it could just be, and often is, just a low spike
With graphs, yes, its maybe more important than average, and certainly highest

Agreed, and using graphs is needed badly as well in benches, as it shows actual fps continuations, instead of simple low/high numbers

Certainly looks that way. Change your CPU clock and see what happens.