Phenom II 955 Versus Core i7 920: Gaming Value Compared

[sighQ2]

[citation][nom]San Pedro[/nom]I'll only add a couple things to this. Someone stated the author said he would choose the i920 for all the other things you do with a computer, but never was this tested in any of the article. Also, hardly anyone really does a whole lot of video / audio encoding in reality (to which I'm sure people on here will say they do it all the time). For the majority of users in everyday computer no difference would be seen.Of course the Intel is faster, we all know that, but I agree with other people saying the cost of the AMD system could have been much lower (940, ddr2, lower cost MB), which is where AMD really is a good buy. In face C2Q is still a good buy too (at least some of the chips). A couple of people said they would like to see results based on one video card as well as the dual video cards tested here, I echo that sentiment. One more thing, there are a lot of people on here who need to take a deep breath, take their favorite relaxant, maybe take a nice long bath, and ease some of that tension/anxiety building up from internet rage.[/citation]
I mostly agree with this post.
except - it is no longer really necessary to overlook AM3, DDR3 = IF BUYING ALL NEW SYS.
If you have EXISTING AM2+ system already, pop in a 940 and you are done.

But, since the price drops, you can now get current AM3 DDR3 and be all new with a future - if you are buying an entire sys w mobo cpu ram.

Apart from that - I agree with you. and I am relaxing more too = good point there also.

 

[sighQ2]

[citation][nom]Why_Me[/nom]So what would you have done different ? I know the Intel could have most likely went to 4.0GHz. The intel had an inferior vid card set up. But what would you do to change up that AMD rig ? Let's hear it.[/citation]

North Bridge oclox = already mentioned. There is a much better way to oclox the P2.

I wish someone would comment on the use of SSE3 - aren't there other versions of SSE?? newer. Do spintel and amd use the same SSE?? I don't think that's quite right - and could be significant to redults - but I really do not know. I wish someone with that info could speak to this issue. I know it's about "instruction sets" and amd writes their own - of necessity. But I cannot comment further - any help thx.

 

[KT_WASP]

[citation][nom]sighQ2[/nom]Is it possible to outline the i7 upgrade path?? I am not familiar with that.[/citation]

I thought I read somewhere that Intel will be focusing on the Core-i5's and i3's in the near future utilizing the Socket LGA-1156. The socket LGA-1366 is going to be reserved just for the top-end Core-i7's. That leaves you to wonder how many more "top-end" Socket 1366 CPUs they will come out with.

I also read that they will make some Core-i7's for the Socket 1156 boards. (AnandTech : "Only the Core i7 is allowed to run in a LGA-1366 socket, however there will be LGA-1156 i7, i5 and i3CPUs . The number of memory channels and the presence of a QPI link does not determine branding. In other words, Lynnfield will be both a Core i7 and a Core i5 depending on the SKU.")

Did I read that right, or am I missing something?

That makes me wonder if Socket-1366 can really be labeled a "good upgrade path". It is not a bad choice, it is just you are starting out at the top already with no place really to go except $500+ CPUs?

Becuase if they really do come out with Core-i7's that are socket 1156 CPUs, then the better "upgradable platform" would be the socket 1156, as you can go from Core-i3 to i5 and finally to i7 all on the same motherboard.

Unfortunately for me,I really don't have any upgrade paths right now. I have a Socket AM2 system (MSI K9A Platinum, Athlon 64x2 5200+ Windsor).. and the motherboard has no BIOS updates for any of the AM2+ or AM3 processors. Hence I will wait a while and see how everything pans out before jumping into a new build.

But I do agree that AMD learned their lesson after the 939 socket change. They lost a bunch of loyal customers when they did that. I think they took that whole episode to heart and are working hard to make sure it is not repeated.

 

[sighQ2]

[citation][nom]Cleeve[/nom]Not true, I'm happy to respond to anyone who is polite.I've heard a lot of complaints about the setup regarding the latency and hypertransport speeds. What all the complainers seem to gloss over the fact that, at best, it might increase the final numbers by a percentage point or two.The irony is, those who left angry comments about imperfect latency and HT speeds know that optimizing the minutia wouldn't make a lick of difference in the big picture. Results and conclusion remain intact.You want me to apologize for not making the bestest, super-duper ultimate AMD setup in the world? Sure, I'm sorry. This was an unscheduled article, I was pressed for time to get it out and I did what I could given the resources I had at my disposal. Maybe not ideal, but let's not pretend it would have changed the parts that count.At the end of the day, even the complainers know it wouldn't have changed anything that matters, especially since the Phenom II ended up with a generous overclock compared to the i7. I don't see you guys complaining about that, by the way. I guess it's easier to complain about the minutia and pretend the overclock discrepancy doesn't exist.I'm a bottom-line oriented kind of guy, real-world builds and a price/performance focus. That's my style. Some of you may hate that; Maybe I'm not the author for you. I can't please everybody and I'm not trying to.If you want to pretend a couple percentage points here or there make a mountain of difference, well, you certainly have the right to think that. But I also have the right to disagree.Bottom line is, I'm still more than satisfied the results of these tests are valid, and support a valid conclusion. To me, that's the important part. The guts of this are solid, and if the wost complaint people can come up with is that they think the latency could be better optimized - well, I couldn't be happier.[/citation]
You have a tough job - and this comparison area will always be a war zone.
I believe we are all a process in becoming. We do the best we can with the information we have. And I see the effort you made for full disclosure - I admit I was surprised to see it on this site.
There are simply a few details - that's all. And we all learn ever more. We start where we are and move forward. I simply want to move forward from this point; because I think the followup to this article could really be something. It's not about perfectionism - that's so self-defeating endlessly. It's about the fact that there is more to this truth - and the driving interest in it is about a relatively small company with 20% marketshare vs a giant abusive monopoly. The abuse story is revealed through the recent (and upcoming) antitrust cases; because that messy ugly stuff has created the marketshare figures. There is more to be seen than just bang-for-the-buck. And it has almost become SO VISIBLE HERE.

I commend you for what you have revealed here. It can go further; and I hope your bust schedule allows that to happen.

I apologize for some of my messy comments - but I had to say what I did, at the same time. There is a greater truth hidden in this amazing display. And I thank you for this opportunity to reveal it; and hope that this can evolve to the next level - that's the truth the world needs to see - and I recognize the financial considerations of following this through - but I daresay, it's about readership - and there is a big win in all this for everyone involved in it.

dude - I have never seen such an attempt at full disclosure - in a war zone. That in itself = I salute you.

You hang in there. Ya done good.

 

[ImaxAMD]

[citation][nom]Cleeve[/nom]Apology accepted.And you should apologize, because if you're telling people that a game will show a 5-10% advantage from running a point or two lower CAS latency - as opposed to the more optimistic overclock the Phenom II enjoyed in this review compared to the i7 - then apologizing is exactly what you should be doing for misleading folks.[/citation]
Its not just about lower latency Cleeve or Don, Its about doing what is right for what you are trying to do, and you dont get it at all, how the hell did you land doing reviews for THG?. Maybe I shouldn't have described it the way I did becsuase you really dont understand do you? I am not saying just low Latency DD2 is better than low CL DDR3, not at all, its just part of the tune, if you have DDR2 or esp if you have DDR3. look at the measely bandwidth you got with that memory, what where those settings again? How You dont answer request for details here nor post them in the review but many came in here and pointed it out, why? Herre is a link for you shows you 5-10% gain in reak use just by good timings, it just touches on it infact as these are mid level chips just like you used..And then there is NB and HT.
http://ixbtlabs.com/articles3/mainboard/phenom-2-ddr3-p2.html

 

[egowhip69]

So basically, at the end of the day, either machine will put up playable framerates... at stock

Now, there is where a bit more research needs to be done. With 1920 x 1200 (or 1080) monitors available for peanuts, we need to have comparisons of different CPU / GPU combinations...

Then we can find the lowest price point to run games with acceptable framerates. To me, THAT is the interest of 99% of the people out there... especially with this crappy economy.

Show the cutoff point of playable at 1920 x 1200. Then, if you want to overclock, tweak, show what $20 more will get you, etc, the more power to you.

I had a friend dump ~$1500 into a build last month. And, yes, he is a spoiled lil rich kid... but even he did alot of thinking about the price / performance ratio before buying his components...

 

[MasterDOOM]

Realy Nice Move AMD - ATI and Great Price My Phenom II 940 Realy Rulz with no OC . AMD - ATI Radeon Best Choice

 

[cleeve]

How You dont answer request for details here nor post them in the review but many came in here and pointed it out, why?

What detail are you asking for exactly, sport?

Herre is a link for you shows you 5-10% gain in reak use just by good timings, it just touches on it infact as these are mid level chips just like you used..And then there is NB and HT.
http://ixbtlabs.com/articles3/mainboard/phenom-2-ddr3-p2.html

Sure. But this is a comparison, so both the i7 and Phenom II get the same timings.

You act as though the i7 results would just sit there statically regardless of the timing changes and the Phenom II would leap a head. That's not realistic, you know better than that. Even though the Phenom II will get relatively more out of it, the i7 will get something, too.

Those results at xbit also show the Phenom II numbers getting a much bigger leap from the overclock than from timing changes.

So considering the Phenom II got a generous overclock compared to the i7, you're saying that if we even things up by lowering the Phenom II overclock (maybe to 3.4 GHz instead of 3.7), tightening the timings (which we would also have to tighten on the i7 to keep the playing field level), and tweaking the HT speeds - you're saying that the Phenom II will get a massive 5-10% boost compared to the i7?

Look at me with a straight face and tell me that you believe that.

Whatever the Phenom II might have lost with loose timings and a low HT, it got back in spades with a generous OC. Come on, man, you're playing devil's advocate when everybody here knows the score.

 

[cleeve]

I an CHALLENGING THG to run the i7 against a little 550BE dual core. You did not cover that - you seem to mostly admit that this might be a REALLY EXCITING REVIEW - but you aren't saying that flat out blatantly.

Neat idea, but a little one sided because of this: I could do the same test with a 550 against a 955 and make it look just as bad from a value standpoint. Cheaper rigs will always deliver more bang for the buck, right? So singling out any high-end rig and pitting a 550 BE against it - well, I can understand folks being annoyed at that.

I'm not saying it would be impossible, just that I'd have to think more to come up with a fair way to play it out.

A long time ago, I actually did something very similar in a system builder marathon: I made a test setup called a 'budget gamer's special', where I pitted an X2 3800+/8800 GTX combo against a C2D E6600/8800 GTS combo. Once we hit 1600x1200 with AA, the cheapo 3800+/8800 GTX won of course even though the AMD setup was more than $200 less! I did it just to prove the point, that games were much more GPU than CPU limited.

But a couple years ago, CPU-limited games were unheard of. Even a cheap CPU could muster over 90 fps in pretty much any title out there. I'm not saying a lot of titles aren't GPU limited now, but I've definitely seen a lot more CPU limited titles in the last couple years.

 

[jab642000]

erdinger 07/13/2009 10:21 AM - you belong reading people magazine not appreciating inciteful information in an everchanging industry with articles written for those of us wanting the best options avail given a specific set of criteria...thanks THG for another great article

 

[sreams]

It is far too easy to skew the results for a desired outcome. The argument that one must choose the Phenom II 955 and DDR3 memory... because the price gap between 2x ATI 4870 and 2x ATI 4870X2 is too great... only works if you assume we -must- start from 2x ATI 4870 in the first place. Many people can't afford to go with a Crossfire rig at all, especially alongside a Core i7 system. How about a Core i7 920 system with DDR3 and a single Radeon 4650 vs a Phenom II X2 550 with DDR3 and 2x ATI 4850 1GB? These systems are -the same price-. Make this comparison, and suddenly your gaming value article looks a whole lot different.

 

[ImaxAMD]

Cleeve having a hard time staying on topic uh?. Others and myself said Timings and Tuning can make a big impact, you said only a “percentage of a point” and latter said it to a point or which betrays your lack of ability and understanding or even the desire to understand, you seem to have no passion for this. You were pointed to some REAL data that shows 5-10%+gained with setup just on timing alone (not even the NB or HT yet) and instead of acknowledging you just try to say "it would have done the same for the I7"...So which is it, you couldn’t tune either system to anywhere near its potential or just the AMD? If there is any review with your name on it, it is valueless until you start showing a bit more zeal or something, you’re a great sport and a decent journalist and decent enough for a review, but you cant come out and say things like you are if don’t REALLY know what you are doing yet. You are much better than I would ever be at this (I would have told everyone to $#$off right or wrong), but please get a tune up on how to set these up, get into some competition or something to brush up. Latter

 

[sreams]

After reading some other comments... The main point here should be: what is the best gaming rig you can get for X dollars? A thorough article would explore more than just one price point. The Core i7 system is only viable as a gaming rig once you reach a particular price threshold, where the buyer can afford two high-end cards in a Crossfire configuration to go with it. Once you get into a price range where only one video card can be afforded for the Core i7, it is too easy to get a far better gaming system with the same money and a much less expensive CPU.

 

[ImaxAMD]

Sorry for my bad english

 

[cleeve]

Cleeve having a hard time staying on topic uh?.

Not at all, I'm quite pleased with how the discussion is going.

As much as you might try and pretend there is no overclock discrepancy in the Phenom II's favor, you can't erase it by ignoring it. And if you're knowledgeable, you know that it affects the results more than a latency and HT adjustment.

You might not like that reality, but I'm not overly concerned with your likes or dislikes, so it's not really a problem for me.

 

[ImaxAMD]

There you have it folks, he again cant even answer for what he said after he was shown some facts from not just from others here but from another ligit review details actual data from another site.

Are you still in high school? I'm sorry, thought this was an adult conversation.
I pretend nothing and supported myself with facts and did not have to change my standing as you have to continue to do.
You on the other hand seem to try to twist things or goad people who disagree with you about how you "set things up" even those who have been polite to you. Why dont you spend more time being concerned about what you do in your "reviews", maybe you will be able to move on instead of having to apologize about them.

 

[cleeve]

I'm sorry, thought this was an adult conversation.

Apology accepted!

I look forward to your next response where you will undoubtedly continue to pretend there is no overclock discrepancy.

 

[San Pedro]

[citation][nom]Cleeve[/nom]I've heard this before, but there's a fundamental problem with this argument.Given the choice between two identically priced rigs, are you saying you would consciously choose the slower one because the benefits of the faster one aren't fully realized yet?Really, how is that a sensible argument? What possible rationalization would back such a decision?Alternatively, I can totally understand the argument that a lower price point could be sufficient; I'm a big fan of the Phenom II X3 720, and frankly if I were going to build a gaming machine for a friend it's probably the way I'd advocate they went. But for argument's sake, if I were to choose between the rigs in this article, I'd go with the i7. I'd choose it even if the 955 had super-perfect latency and HT overclock settings that gave it a 2% performance increase over these results (blasphemy! I must be an AMD hater! ). I don't think that's bias, I think it simply makes sense.[/citation]

Of course at the same price point get the i7, BUT you could have done an AMD build much cheaper with pretty much the same performance. It does seem like you stretched the budget on the AMD build to match that of the i7. . . However, I must say that this was probably the point of the article, comparing an AMD system with an Intel system at similar price points, but I think it would have been better to do a non-CF test for both systems as comparison.

I think people would have liked to have seen a cheaper PII 940, or x3 720, with DDR2 and then even out the price difference with the graphics subsystem, so maybe you can get a AMD build with two graphics card, but for the same price you only get one graphics card with Intel. That would be much more relevant to people looking to buy a gaming machine. (Also should mention I'd still want to see both systems tested with just the one card a piece.)

 

[playerone]

Apology accepted!

I look forward to your next response where you will undoubtedly continue to pretend there is no overclock discrepancy.

A few of us are gathered here and watching for fun...

I think he called you out three times and what you stated, he did back up his claim with a link and you are the one that pretended.

I'll point out he never mention or discussed O/C discrepancy, but now that you bring it up:

$170 940s are good for 3.8-3.9, but you spent the money on a $210 955 and clocked it at 3.7??? 955 are good for 3.9-4.2 on air WTF? I7 was DDR3 so you had to go with the DDR3? Oh, ok. but then why did you set it up with less bandwidth than most DDR2 can be pushed? and then again why did you only do 3.7?

I supose you will now change the subject...

 

[ImaxAMD]

[citation][nom]Cleeve[/nom]Snip... there is no overclock discrepancy in the Phenom II's favor, you can't erase it by ignoring it. And if you're knowledgeable, you know that it affects the results more than a latency and HT adjustment.snip...[/citation]

I'll erase your childish emotive comments and try to keep it objective for you...

So you are saying the CPU O/C is more important than the Dram timings, HT, NB subsystem? The is that why you held back a 955 to 100mhz less than any 940 can do? Are you saying you held both of these systems back? by how much? I Just read your other review "CyberPwr" very intresting, How long will it be before we can expect another great review and will therer be an opology in there for this one too?

 

[noob2222]

[citation][nom]imaxamd[/nom]Its not just about lower latency Cleeve or Don, Its about doing what is right for what you are trying to do, and you dont get it at all, how the hell did you land doing reviews for THG?. Maybe I shouldn't have described it the way I did becsuase you really dont understand do you? I am not saying just low Latency DD2 is better than low CL DDR3, not at all, its just part of the tune, if you have DDR2 or esp if you have DDR3. look at the measely bandwidth you got with that memory, what where those settings again? How You dont answer request for details here nor post them in the review but many came in here and pointed it out, why? Herre is a link for you shows you 5-10% gain in reak use just by good timings, it just touches on it infact as these are mid level chips just like you used..And then there is NB and HT. http://ixbtlabs.com/articles3/main [...] r3-p2.html[/citation]

Is it just me or am I totally comfused .... saying memory timings is very important, but this article only covers DDR2 vs DDR3. and from what I see, ddr3 is faster. Who woulda ever thought of it.

here is my link http://www.tomshardware.com/reviews/phenom-ii-ddr3,2319-6.html

imagine that, in games, faster memory = faster fraps with the exception of UT3 wich gains .1 fps with low latency memory (little math here, 134.2/134.1, so I can gain 0.08% faster FPS with lower latency in one game OMG I am gonna spend weeks trying to get that maximum >1% tweaked out of my system.

Just to point out, the 955 memory overclock is faster than the i7 in case you missed that also.

If you are referring to a 5-10% gain from your link, its from the 5-10% overclock from the CPU (noted at the top of the chart, 2.6 oc 3.7.

anyway, nothing personal, just in a bad mood today, felt like argueing about something

 

[pauldh]

[citation][nom]sreams[/nom]It is far too easy to skew the results for a desired outcome. The argument that one must choose the Phenom II 955 and DDR3 memory... because the price gap between 2x ATI 4870 and 2x ATI 4870X2 is too great... only works if you assume we -must- start from 2x ATI 4870 in the first place. Many people can't afford to go with a Crossfire rig at all, especially alongside a Core i7 system. How about a Core i7 920 system with DDR3 and a single Radeon 4650 vs a Phenom II X2 550 with DDR3 and 2x ATI 4850 1GB? These systems are -the same price-. Make this comparison, and suddenly your gaming value article looks a whole lot different.[/citation]
I think you are missing the point to this article. It was a follow-up to another article using the PII 955 and Dual graphics cards. The author addresses the main reader concerns with that article. You need to take it for what it is and not try to reinvent it into another story. There is nothing wrong with using AMD's top CPU vs the i7 920. And the twist of different graphics cards to equal budget separates it from a typical CPU review. Sure, both solutions may be too costly for you and I, but that doesn't take away from it's value.

In your example, the i7 is ridiculously unbalanced for a gaming rig. I'll add a third system though. To challenge you PII 550 + HD 4850, let's put in an E5200 + HD 4870. Max overclocked, guess which system will have the most game? I'd gladly do that comparison (and actually put that request in many weeks ago), but in the end the results wouldn't directly add nor take away from this story that uses higher end components.

To others: the author has not made claim this is a fully tweaked AMD setup. Can't you see that pushing voltages and getting 3.8GHz on the CPU, 2.7GHz on the NB, and running tighter mem timings is not going to gain more performance % than adding 500MHz more to the 920's OC? Had this been a 4.0GHz i7, then you'd have a valid beef on the PII's OC. Would you have been happier if the PII was tweaked out on NB and latency, but left at stock VID and running 3.4 GHz? I'm glad the majority of readers actually understand/appreciate the comparison being made rather than read solely to find fault.

 

[Crashman]

[citation][nom]pauldh[/nom] To others: the author has not made claim this is a fully tweaked AMD setup. Can't you see that pushing voltages and getting 3.8GHz on the CPU, 2.7GHz on the NB, and running tighter mem timings is not going to gain more performance % than adding 500MHz more to the 920's OC? Had this been a 4.0GHz i7, then you'd have a valid beef on the PII's OC. Would you have been happier if the PII was tweaked out on NB and latency, but left at stock VID and running 3.4 GHz? I'm glad the majority of readers actually understand/appreciate the comparison being made rather than read solely to find fault.[/citation]

Those complaints about the AMD overclock are a scam. Nothing more than lies and deception made to tell the exact opposite of the truth.

Everyone can see that the AMD overclock was "better" than the Intel overclock, in that the AMD overclock was far closer to max than Intel. If anything, Don cheated against Intel on this by giving AMD a better overclock.

It's like affirmative action, you have to intentionally handicap the majority to keep the minority from screaming foul. And no matter how much you handicap the majority, if the minority fails to overwhelm they still claim discrimination.

Let me repeat that in words everyone understands: Don discriminated in favor of AMD, yet some of the AMD faithful are still screaming that he didn't favor AMD enough.

It's all a big scam by a few of the "black helicopter crowd" to discredit anything close to the truth.

 

[playerone]

[citation][nom]noob2222[/nom]Is it just me or am I totally comfused .... saying memory timings is very important, but this article only covers DDR2 vs DDR3. and from what I see, ddr3 is faster. Who woulda ever thought of it.here is my link http://www.tomshardware.com/review [...] 319-6.htmlimagine that, in games, faster memory = faster fraps with the exception of UT3 wich gains .1 fps with low latency memory (little math here, 134.2/134.1, so I can gain 0.08% faster FPS with lower latency in one game OMG I am gonna spend weeks trying to get that maximum >1% tweaked out of my system. Just to point out, the 955 memory overclock is faster than the i7 in case you missed that also. If you are referring to a 5-10% gain from your link, its from the 5-10% overclock from the CPU (noted at the top of the chart, 2.6 oc 3.7. anyway, nothing personal, just in a bad mood today, felt like argueing about something[/citation]
Hi,
If you go back and read the post and the link I think the point was to look at the chart where its the same over clock and where they discuss the impact of Bandwidth, Command rate with the DR2 and DR3, same CPU clock, different memory and timings = %10 at the same CPU overclock. But you dont have to read a review to figure that out, benchmarking a phenomII is different than an I7 and that was the point, the I7 loves the memory just climbing away at speed, so does the PHII, but its different balance of speed, low timings and command rate because ata certain speed the HT has to be set back because if you dont you will see performance start to fall off. There is a falacy abouth the bandwidth numbers, dont look at those too much, or at least if you do test the cpu strength at the same time and where the two paths cross thats where you want to be. I think Don's angle was great but it falls off the tracks in execution. its not about I7 vs PII,

 

[ImaxAMD]

[citation][nom]noob2222[/nom]Is it just me or am I totally comfused .... from what I see, ddr3 is faster. Who woulda ever thought of it.here is my link http://www.tomshardware.com/review [...] 319-6.htmlimagine that, in games, faster memory = faster ...anyway, nothing personal, just in a bad mood today, felt like argueing about something[/citation]
You are using those benchmarks? uhmm you might rather use 1066 or 1200 at 5-5-5-15, would like to compare those figures and see what blows what out of the water? You run 1066 at 8-8-8? ZZzzzzzzzz Zzzzzzzz
Look, $50-60 will get you 4-4-4 to 5-5-5 kit, $120+ can get you a 8gig kit that runs 1100 at 5-5-5 at just 1.8v hmm what happens when we go to modest 2v? 12-1300 easy at super quick timings. Just look up G.skill on newegg or OCZ or what ever.