Discussion in 'Article Discussion' started by CardJoe, 7 Dec 2010.
switching from a 965 to an i7 950 is one of the best upgrade I have ever done ... I don't recommend an AMD cpu for any gamers out there, they suck at it.
Hmmm the scores in the Bit tech benchmarking suite (which is more reprensentative of real world use) are very dissapointing with this CPU at the stock frequency of 3.3Ghz. It is only ~40% faster than a 2.66ghz dual core E6750....
Rather more telling is the fact that when overclocked it is barely faster than a overclocked G9650 (proxy for Core i3) which is significantly cheaper.
It does do better in the multi threaded Cinebench but then it is still behind the core i7's.
Gaming performance looks ok but TBH more tests need to be done before any conclusions can be drawn.
Overall I am dissapointed. There will be occassions where this CPU will be brilliant but most of the time it seems to just be run of the mill and generally outclassed by the Core i5 and i7 models from Intel.
AMD really need to step up its game, at this rate, we might see P67 platform staying around forever.
With intel being socket happy at the moment, i don't think there's any danger of this happening. It would be nice for intel to have some competition to force lower prices though.
Utterly ridiculous generalisation.......check any number of sites with benchmarks, and you will see the Phenom II CPU's can hold their own in any game out there...
Your graphics card is the single biggest factor with games today, not CPU's with more than 2 cores.
Jeez, even in this review the 1100 is only about 10% slower than the 750.....if thats your definition of 'sucks', then you need to get out more.
To be fair, 10% slower and potentially more expensive does kind of suck!
"Our multi-tasking test performs a massive file backup (with encryption) using 7-Zip, while simultaneously playing back a HD movie file using mplayer, making it a demanding test for any PC."
Don't get me wrong i am not critising the review but a score of 1038 for a 6 core cpu @ 3.3ghz vs a core2 @ 2.66ghz in a multitasking benchmark is terrible! Is that really representative?
I admit I am not the average computer user but i would guess that the majority of people that read this website would have more than multitasking than those two tasks. For example i would have video encoding, backup + encryption, a virtual machine running and maybe some light internet browsing going at the same time.
Otherwise excellent review, i was debating the phenom vs i5 for an upgrade but the only reason i was even considering the phenom was the six core and the price. I guess i can wait till sandy bridge then.....
That's the last Bit-tech CPU review I'm going to read.
Seriously guys, we used to have some of the most in-depth and respected testing methodologies, evident in every review.
- When did transparency and fairness in test setups get thrown out of the window? e.g: what ram timings and speed were used for each setup?
- Who decided that discussing performance at stock speeds, and therefore comparing products based on price/performance and value was unnecessary? e.g: discounting overclocked results, AMD's products are much stronger. There's also no mention of comparative prices between the products you have tested against.
Bit-tech used to be run by enthusiasts, for enthusiasts.
- Where's the excitement, and passion for the industry gone?
The way things are, I feel embarrassed to have worked for the site.
I've got the 1090T, and i'm slightly confused as to why you have upped the voltage so much on the cpu/nb and on the actual NB and SB. Mine atm is running at 4250mhz with 1.5375V through the core the cpu/nb is at 1.2 and the nb and sb are at stock voltage. Obviously all the chips are different, and if you were running tighter ram timings that would push up the cpu/nb, but why put up the nb and sb? As it is a BE why not just up the multiplier more to take less stress off the board?
Fan boy alert!
Unfortunately, AMD couldn't supply us with an accurate UK price before the launch, so we've made a best guess by comparing the 1100T BE to the Core i5-760 and Core i7-930.
However, to address your other points, the memory frequencies and latencies are listed on page 2, titled Test Setup. We compared stock-speed performance on page 8, titled Performance Analysis and Conclusion.
Frankly, with reading skills like that, I'm embarrassed you used to work for the site...
Thing is, nobody with half a brain would buy a 6-core CPU for gaming. Buy a dirt cheap 3 or 4-core AMD CPU and you'll be laughing at everyone who buys Intel instead.
The problem is CPUs and GPUs are always reviewed seperately. Yes, if you take a specific graphics card and test it with an i7 920 vs. a Phenom II X4 you will get better framerates with the Intel. If you take the price difference between the two (over £100) and spent that money on a better graphics card for the system you'll see a HUGE increase in gaming performance. Two systems with the same overall cost, one much quicker than the other. I know which one I'd choose and it wouldn't be the Intel, sorry.
That's actually a very good point, but I also suspect it's nigh on impossible to review like that as the combinations required to cover all bases would be frightening - it'd probably take a fortnight to review every single new GPU / CPU.
I think it'd be a very good idea for a one-off article though: Say a set budget of 'x' pounds, looking for the best combination of MoBo, CPU & GPU - all other components identical...
The Crosshair IV Formula we use to test we use to test AMD chips is an awesome overclocker. It can handle a HTT all the way up to 350 so we overclock on the HTT rather than the multiplier as this gives more CPU to system bandwidth.
Having the HTT run so quick puts extra stress on the Northbridge (and to a lesser degree the Southbridge), so we whack a little more voltage through them just to make sure they are happy.
It varies from game to game too, depending on how well multi-threaded they are. Games that only use one core run much better on a heavily overclocked Intel chip whereas games that use lots of cores run much better on AMD offerings at the same (or lower) price point, simply because they tend to have more cores available.
Supreme Commander is a great example. When it came out it needed the fastest Intel you could afford overclocked as far as you could overclock it. Now that it's been modded to run on multiple cores it runs as well or better on an el-cheapo AMD CPU for significantly less cash.
Bit-tech: I'd be interested to see a multi-core supcom benchmark for the gaming section of CPU reviews. You could even run it without the multi-core mod to test single-core performance if you wanted. I imagine this X6 chip would do rather well in the former but suck royally at the latter.
I never talk out of my ass. When I say something sucks it's because I've tried it enough to draw my own conclusion.
I used to have a 965 overclocked to 3.8Ghz and couldn't have acceptable (over 85) framerate in Black ops (I know it's not a reference), TF2 (now that's not normal), L4D2 and BC2. I tried my friend's GTX470 and saw no improvement whatsoever. I also tried 2 GTX260 in SLI and saw no improvement either. Oh yeah that's right, I even tried my friend's 1090T on my board and it was slower than my 965 because, for some reason, the new bios to support the 6core would not apply the overclocking settings. As mentioned in the thread I linked to, the cards (whether they were in SLI or not) were never over 50% of usage
Two weeks ago I had an opportunity to upgrade for really cheap so I got myself a 950, 3x2GB of A-Data PC16000 and a Sabertooth X58. Before I overclocked the cpu, I tried TF2 and was blown away... I now average between 110 and 160fps with peaks at 300 whereas I would average 50 and peak at 80 with the phenom. I agree, those numbers look like they are definitely exaggerated but it's a rue story ... I couldn't believe it either. The same thing happened in the other games as well. My games are now more fluid than they are on my friend's pc (1090T @ 4.1Ghz and GTX470) and my 260 is pretty much always used at 95% and over
Let's recap, shall we ?
The 965 @ 3.8Ghz bottlenecks a 260
The 965 @ 3.8Ghz bottlenecks a 470
The 965 @ 3.8Ghz obviously bottlenecks a SLI of 260s
The 1090T @ stock bottlenecks a 260
I think I have the right to believe that phenoms suck for gaming
Your move, buddy
I feel like I need to clarify something before everyone calls me a god damn fan boy. I am not a fanboy of either companies (the same applies to ati and nvidia), I just want what's best for my needs and, clearly, AMD isn't for gaming.
Nice one Knuck. I actually think you have every right to call AMD whatever the hell you like. Like you, I have also spent vast amounts of money on both Intel based and AMD based rigs. The current standing and for quite a few years now, I really haven't seen anything of any merit from AMD to warrant even the most hardcore fanboy defence.
But still they go on....and on....and on..... and on about how wrong we all must be when most of them are probably still using an AMD64!
And the 750 is cheaper than the 1100... Yep, that's fall under my definition of "sucks".
EDIT: The "sucks" above was referring to if it was a new build (although you could get a G6950 and mobo for the same price as this CPU, so that may make a better upgrade (if you wanted a system for games(according to the game performance charts in the review)) dependent on a gratuitous number of other "things"). It's only because I've had some not-as-bad-as-bad-but-a-bit-below-normal experiences with some AMD processors I've had in the past.
Separate names with a comma.