i dont like it - THG makes it out that that Intel 920 strugles to keep up with the AMD and is a bit behind, Anandtech makes it out that the AMD is no match to Intels might even with its base model - which is it?
Ask intel or amd. They're the ones that may know. Everyone else, including thg can only make qualified guesses based on their testing.
You don't have to agree with either of those two sites you mention, but ultimately you'll have to pick for yourself whom you want to trust if you are convinced you cannot trust both.
Well from all other sites an Intel Q9550 is around the P2 940 mark and the Intel i7 920 blows that chip out of the water so im sticking to Intel here as the current champ regardless of model - there all quicker.
I've been running benchmarks all day long specifically in response to this discussion here today. I was hoping to be done tonight, but I have at least one platform left to test. I'll have more for you guys tomorrow, but the bottom line here is: the numbers in this piece are correct, and I have about eight possible scenarios worth of troubleshooting with supporting data to support them. Thanks for being patient guys--I'm just as intrigued by the differences being reported as everyone else.
[citation][nom]VTOLfreak[/nom]I wonder if the low scores for i7 may have anything to do with sample CPUs, pre-production motherboards and beta BIOS'es. I doubt THG buys all their gear retail like we do.The missing memory multipliers on sample i7's prove that what review sites use and what actually ends up on the shelf are not always thesame.[/citation]
This is true--though we do have several retail 920s that've come through the lab for our System Builder Marathons. Performance is comparable on the retail and ES chips if you're using a DDR3-1066 multiplier. Naturally, the retail sample will let you go higher.
Can the same setup be tried slowing the mem down to min and making timings as tight as possible on the i7? I have been doing a study on the real (non-synthetic) benefits of timings v/s high clocks for user/gamer level operation as games rely more on computation (mem reads/writes) than raw bandwidth?
this number is clearly screwed. sure we can do our own benchmark. But THG does this as a commercial website. Don't they have to be more professional. so stop your lame comments and just answer technical questions technically: why GPU score is of so much difference for the same card?
[citation][nom]rdawise[/nom]Ok unforunately for the author he didn't say that Intel is still the God of all CPUs so he has come under fire...for shame. Listen guys if you find their results so question why don't you run out, run your on benchmark test on a similar setup, then post. Stop the fan boyism and read the article for a minute. When you get your benchmarks up to contradict, then say something. As for the gentleman who posted the links to different articles, there are some major differences that you are neglecting. From both articles they used resolutions that were lower than 1680. THG bottom was 1920. Also, look anandtech's memory timimngs versus THG again different. Also different mobos on anandtech. But the biggest difference I think neglected here is the fact that THG was the only one (stated at least) that used an AM3 mobo versus the AM2. Could those be the reasons you see differences...I would be inclined to think so. Anyone here have a AM3 mobo and a PII that can post benchies that can negate these listed here? I thought not. Chris great article, sorry for some of our peers here.[/citation]
I feel bad for people who bought I920 and now are pissed because it can't even beat a tri core. Sorry guys but AMD is making a comeback. There are other sites out there than anandtech and they do correlate with Toms on this one! Here is a post from someone on Anandtech!
thanks for the links which are interesting to read but sorry your links are not saying what you said.
Just look at 3DMark Vantage in these various reviews, you will see how i7 is leading the pack, although margin varies, and it is not like 20% falling back as it is in this review.
Then you need to understand the benchmarks. When CPU is limited in low res, you see clear benefit of i7. When GPU is limited, all numbers are close enough to be in the error margin. 790 maybe is a very good gaming chipset that gives COD4 some x% "extra boost" at very high resolution. But that is about it, and at high resolution, current/old games just don't care CPU that much, which is very true.
[citation][nom]beeyang78[/nom]I feel bad for people who bought I920 and now are pissed because it can't even beat a tri core. Sorry guys but AMD is making a comeback. There are other sites out there than anandtech and they do correlate with Toms on this one! Here is a post from someone on Anandtech!I think anandtech is intel's doll.refer to this reviewshttp://www.overclockersclub.com/re [...] 0_810_am3/http://www.tomshardware.com/review [...] 48-10.htmlhttp://www.guru3d.com/article/amd- [...] iew-am3/22http://www.guru3d.com/article/amd- [...] ew-test/18Core i7 is absolutely a gaming failer!![/citation]
These benchmarks align with the initial Core i7 benchmarks, which indicated it sucked at gaming. Then, it magically started kicking ass at gaming for no apparent reason, and nobody questioned it. I would imagine that Tom's and Intel might have had a falling out, hopefully this means no more epic fail articles like:
Core i7 released: OC the new CPU, but not the competing CPUs, on the hopes that people will look at the graphs and not read the article.
Core i7 reviews: Make an unusually high number of the benchmarks play to i7's strengths, synthetic benchmarks and video encoding, never before had the benchmarks been so focused on these 2 areas, avoid discussing how Radeon GPU video encoding wipes the floor with ANY cpu.
Phenom II vs. i7 OC: Use crappy mobo+heatsink for the PhII to cripple the OC, which shifts numerous close victories to the Core i7.
That obsolete argument of the benefits of certain cpu's al low res, who cares? No the processor A is better at gamer resolution in 1 year it will be the same. In two years gamers will upgrade/change their rigs. The circle will restart over and over again.
You need to care about the performance now, A option is better than option B will less money? Don't like it? Bear with it.
All I am saying is a normal 3DMark will give the similar GPU score for a relatively fast system with the same video card. It is abnormal for 3Dmark gives a 20% lower GPU socre for a CPU has faster CPU score. This observation has been shown many many times by almost all, even by THG's own test in Jan09. If you don't understand that, you may do your own testing or take a computer architecture course.
[citation][nom]Nintendork[/nom]kknd1967Do you play with 3Dmark? Is it funny? NOThat obsolete argument of the benefits of certain cpu's al low res, who cares? No the processor A is better at gamer resolution in 1 year it will be the same. In two years gamers will upgrade/change their rigs. The circle will restart over and over again.You need to care about the performance now, A option is better than option B will less money? Don't like it? Bear with it.Or do your damage control.[/citation]
Is disappointing that we can't have faster AM3 chips. Also disappointing that AM3 does not support triple channel (Why not quad channel?).
Not sure, but Phenom II allready support ddr2 and ddr3 in the same package, so It may be one reson why (only) dual channel.
Other explanation can be that they find out that there is very little difference between dual and triple channel (same thing as with i7)
In anyway AMD have much tighter budjet than Intel, so it does emphasis things that are most cost effective (as long as they can...) So the abilyty to produce prosessor that support two different memory systems is cheaper to them than making two different CPUs. But as allways there are compromises to be maid. It can be so that AMD goes to triple or quad channel when they can make 8 cores (or more) chips that really can benefit for triple channel, and when the ddr3 is cheap enough, so they only have to support that memory type.
For AMD CPU and MO compatibility:
1. AM2 CPU can be used on AM2 and AM2+ MO;
2. AM2+ CPU can be used on AM2(if BIOS support) and AM2+ MO;
3. AM3 CPU can be used on AM2, AM2+ and AM3 MO, AM2 and AM2+ MO also require BIOS support.
It's easier for MO manufacturers to sell stocks, also esay for us to upgrade CPU.(If you think upgrading is worthy) While Intel is going to use 4 different sockets for new generation CPU，This will make same mistake as AMD did in 754 and 939.
[citation][nom]aganainotoki[/nom]For AMD CPU and MO compatibility:1. AM2 CPU can be used on AM2 and AM2+ MO;2. AM2+ CPU can be used on AM2(if BIOS support) and AM2+ MO;3. AM3 CPU can be used on AM2, AM2+ and AM3 MO, AM2 and AM2+ MO also require BIOS support.[/citation]
correct, and some slacking manufacturers won't bother updating bios support. For instance biostar on that board I mentioned earlier.
If they updated these charts whenever they do an article like this they would always have the 'MOTHER OF ALL CPU CHARTS'. Sure it would take some work getting these charts caught up to current times, but then it would only take minutes when new processors come out.
I have seen these charts posted on the walls of several computer repair shops although they haven't been updated in years. For those of us that don't have a photographic memory, these charts really come in handy sometimes.
I think something is wrong with the 'add a url' button. I copied and pasted the address into that button and they didn't show up either time that I pasted them. I just pasted the address as plain text into the box and it shows up as a link.
[citation][nom]jameskangster[/nom]I apologize for keep posting other site's reviews, but it just bugs me that the other site's have different results posted compared to Tom's Hardware. Extremetech's review (http://www.extremetech.com/article [...] 569,00.asp) benchmarks are also similar to Anandtech's results. Although, they use Nvidia 9800 GTX. Maybe Tom's Hardware could review its performance tests and systems setup?[/citation]
Sorry, but read the article. extremetech "We really wanted to check out performance of the 810 on a DDR3 platform, but as we mentioned earlier, our sole DDR3-capable AMD board was DOA. We'll redo all these tests on faster DDR3 memory at a later date"
Long story short-AMD PIIx4 940 with ddr2 1066 & 790gx/fx-750sb hangs with or beats i7 x58 w/ ddr3 in real world performance. Which do you think is a better cost/performance ratio? What if you already had an am2/am2+ MB and ram? Have fun building a complete new system Intel fans. The AMD crowd is laughing all the way to the bank.