Early Ivy Bridge Benchmark: Graphics Performance Greatly Improved
The folks over at Anandtech managed to spend some time with early Ivy bridge production samples and perform a few benchmarks. The skinny: CPU performance is mildly increased as expected, but the GPU is 20-50% faster than the Sandy Bridge GPU. Power consumption is also down about 30W under full load. The graphics, however, are still slower than AMD's Llano (but the Ivy Bridge CPU beats the pants off of the Fusion's). Is the tradeoff worth it?
It isn't meant to be powerful graphics. It isn't a "tradeoff". Intel's HD graphics are meant to be very low power, but competent enough to run basics, shiny OS features at least. That they do, and it sounds like IB is even better at that. But it isn't a "tradeoff" to get a good CPU with basic graphics that is called "normal". If you need good graphics discrete is still the way to go and there are plenty of reasonable options.
From the look of it, Ivy Bridge is quite a win. Sandy Bridge, but a bit better. Nothing not to like there.
Frankly, I am sick and tired of these integrated GPUs. The theory is that its a cost saver, but since I just put in a dedicated graphics card it ends up being a cost with no benefit. Ah well.
"Have you ever thought about just turning off the TV, sitting down with your kids, and hitting them?"
Does GPU performance matter? Are people using integrated graphics for graphics intensive things? It seems to me that integrated graphics performance has more than exceeded that required for normal desktop graphics for a while and anyone who is doing anything seriously graphically intensive is using a discrete graphics card. Am I wrong?
So basically all laptops that have discrete graphics have it socketed in an nVidia MXM slot. Way cheaper to have one board and just knock cards on it for the manufacturers. However the thing is that since it is for OEMs and not consumers, it isn't as easy to swap as a PCI card. It is all on you to make sure the card you are getting is physically the right size, electrically something you system can handle, and thermally not to much.
Also pretty much only Sager actually supports doing it, and other laptop manufacturer will tell you to GTFO if you ask them about it. As such even finding the parts isn't easy.
With laptops you don't really upgrade much other than maybe the RAM or disk.
However the IB will be useful in laptops not only because it can give better performance for integrated only systems, but it'll be nice for switchable ones. You can get ATi card systems where you can manually switch between discrete and integrated and nVidia ones that do it on the fly. Better integrated graphics means you can use them for more things, so when on battery it is more feasible to use them and leave the discrete system shut down.
However note this wasn't a laptop part they are talking about, this is the desktop part.
If you do office type work only on a machine, then up to, say ~20% performance differences are pretty much irrelevant CPU-wise, as long as you're in a certain performance class. At least as important is harddisk performance, and having enough RAM to do jobs properly.
With any more-than-casual gaming, chances are GPU performance is much, much more important than CPU. So a small CPU advantage would still lose if the GPU is weak. Wouldn't know about their latest, but historically performance of Intel integrated GPU's has been pathetic compared to Nvidia or ATI (AMD) counterparts.
And there is a lot of use for them.
In terms of desktop chips it is for low end use. A lot of people just do web/e-mail/word with their systems and an Intel HD graphics setup is perfect for them. It is plenty of power to do the shiny OS interface, accelerate video, and so on, and comes with the system.
In terms of laptop chips, you really always want it on account of switchable graphics. If your laptop has switchable graphics it can use the integrated for low power consumption and only kick on the discrete when needed. For ATi cards it is a bit clunky, you have to actually manually switch it, but you can do that just use integrated on battery, discrete on power. For nVidia they have a thing called Optimus where it all happens in realtime without you noticing. You'll be on integrated on the desktop then you fire up something intensive and it switches over seamlessly.
Regardless, they are widely used and so worth including. It would cost more for Intel to make a second variant of the chip without them.
Even an ION board blows the doors off a Voodoo from a decade ago.
I've noticed that Best Buy doesn't have the ION-based Aspire Revo PC anymore. Is the EeeBox any good for those who choose to buy rather than build? And I thought NVIDIA had got out of the Intel chipset business anyway due to patent licensing squabbles.
How does ivy bridge compare to ARM? From what I've read, it appears that the ARM has lower wattage, but I'm not sure about the performance.
The CPU in Llano is 2 generations back... with Athlon II. Beating the pants off Bulldozer is easy for Intel: just find a benchmark optimized for single threads, compiled with ICC, or weights the single threaded result. One of the major new features, the random number generator, wasn't even tested. Monte Carlo benchmarks, where are you?
http://www.anandtech.com/show/5626/ivy-bridge-preview-core-i7-3770k/11
It was faster than low end cheapo cards. Which is mainly the point.
If you are putting in $200 cards, they are a long ways off, but they essentially obsolete the need for a low end card, which is a good thing.
And since all most people need is a low end card, this is sufficient for most people.
For desktop, internet, video, web games, older games and even new games at modest settings this is fine.
There will be more off-loading of specific types of tasks as GPGPU becomes more mainstream. One of the things that will be driving wider adoption of GPGPU is better support in tooling. The MS supported extension to C++ called AMP should enable many more developers to take advantange of DX11 (or better) hardware for non-graphics work. It looks like support for AMP will be in Dev 11. If Ivy Bridge has DX11 support with more than a few GPU cores, then Ivy Bridge users should bennefit.
It depends on what you're doing with it! Duh... Seriously, that was a deeply stupid question.
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
1. AMD CPU bug
2. AMD divesting from its fab
3. Intel pulling even MORE ahead on performance and even lowering power usage at the same time!
Not to mention AMD's financial troubles and the fact they have a tendency to burn up.
Just because it CAN be done, doesn't mean it should!
I'm a bit confused by the target market for these improvements. If you're buying one of these fancy chips for a desktop, you must have some reason to need all that power, and 90% of the people who have such a reason will also need their computer to have a discrete graphics card. If all you're doing in Facebook and photos, a cheap Core2 duo is more than you need. If you're gaming, you still can't do it without a discrete card. So now we hear that the only thing that really got improved in this generation is the integrated graphics, the one feature that just about no $250+ CPU buyer uses anyway. It would be one thing if the graphics portion could be repurposed to run some vector commands which the CPU could offload - but nothing like that appears to be in the works. So it seems to me that the graphics is just a waste of die space. I'd much rather see that used for CPU cache. Seriously, how hard would this be for Intel to do? It's not like a mask with extra cache instead of GPU would be all that hard for them to design!
Soooo, you built a CPU that barely runs faster than the previous generation CPU. However the integrated graphics are 20-50% better.
Integrated graphics for anything other than the most basic tasks are horrible by several degrees of magnitude. You can buy a 130$ discrete video card that will deliver 1000% time graphics.
In real world terms this is like taking a game that runs at say 12FPS and making it run at 14-18 FPS which is still unplayable. More realistically you will take a game that is completely unplayable because the graphics won't even run, to making it barely run, but still being unplayable.
The 30W less of heat is more interesting, may make for a better overclock or at least safer.
I guess this does advance the bar for integrated graphics, which academically is good. Realistically however no one will care for years, as the people that upgrade to the newest hardware are also the same people NOT using integrated graphics at all.
Good I suppose for that laptop market, though I didn't see a distinction here between desktop and mobile versions of the chip, so I assume they tested the desktop version.
As for the AMD quip, quit being stupid. So you buy a CPU solution that has slightly better integrated GPU but gets its ass handed to it in CPU? Makes sense.
Crack them open some time. Slots are the big thing since it keeps production costs down.
The question is do you want to game, and at what price... Because for excellent prices you can buy quad core AMD Llano laptops with all the CPU performance you really need and the GPU performance to play a lot of games. Why would you want to go the Intel route, pay more, and get less gaming performance?
"but money is the God of Algiers & Mahomet their prophet." - Rich. O'Bryen June 8th 1786
So how exactly are you using these templates? Could you achieve the same by using a preprocessor?
If indie developers should be developing for smartphones, not consoles or Ivy Bridge PCs, then how should the input method work? Specifically, how are platformers and fighting games controlled on handheld devices running Android or iOS, which rely on a capacitive multitouch screen for input and have no tactile feedback for which buttons the player's thumbs happen to be over?