Second-Generation Ultrabooks: Faster And Cheaper With Ivy Bridge

Benchmark Results: WinRAR

WinRAR is one of three archival apps normally included in our processor reviews. Historically, it took far better advantage of multi-core CPUs than WinZip, though Corel's recent 16.5 release improves that app's threading. WinRAR doesn't always scale as well as 7-Zip, though.

This title isn't without its quirks, though. For instance, the system built around Intel's 17 W Core i5-3427U predictably trails the 45 W Core i7-3720QM by 30% (~23 seconds). However, the Core i5-2467M, also a 17 W chip, surprisingly beats the Core i7-2820QM. In every way specification the i7 outclasses the i5, suggesting something wrong on our end. But we repeated the test several times to the same result.

We've never seen WinRAR achieve high processor utilization numbers. It's tempting to suggest a storage bottleneck, but even when we run this test from a RAM disk, CPUs tend to flounder about in the 50% range. Regardless, the outcome in WinRAR is similar to our Photoshop metric, with the quad-core chips tending to hover between 20 and 30% duty cycles, as the dual-core models are forced to work harder.

Both of the 17 W parts behave as we'd expect when we turn to power consumption, sipping energy until their respective workloads are finished. The 45 W models exhibit vastly different tendencies, though. The Ivy Bridge-based Core i7 platform spikes up, hurries through its task, and dips back down. Meanwhile, the chip employing Sandy Bridge maintains more conservative power use. It's punished for its moderation by a far worse performance result.

Create a new thread in the US Reviews comments forum about this subject
This thread is closed for comments
31 comments
    Your comment
    Top Comments
  • It seems that Ivy Bridge's lower TDP and its HD 4000 comes into its own in the notebook/ultrabook market more so than the PC/gaming one.
    12
  • crisan_tiberiuthey should use HD 4000 on every intel CPU, i dont get it why they dont ...


    Pricing, TDP, segmentation and PROFIT
    10
  • Other Comments
  • It seems that Ivy Bridge's lower TDP and its HD 4000 comes into its own in the notebook/ultrabook market more so than the PC/gaming one.
    12
  • ^agreed

    Andrew Ku
    Tom's Hardware
    6
  • they should use HD 4000 on every intel CPU, i dont get it why they dont ...
    -3
  • crisan_tiberiuthey should use HD 4000 on every intel CPU, i dont get it why they dont ...


    The integrated graphics is built into the processor die and the changes between the HD 3000 and 4000 are physical changes, so they can't just change them without changing the whole CPU.
    3
  • crisan_tiberiuthey should use HD 4000 on every intel CPU, i dont get it why they dont ...


    Pricing, TDP, segmentation and PROFIT
    10
  • sam_fisherThe integrated graphics is built into the processor die and the changes between the HD 3000 and 4000 are physical changes, so they can't just change them without changing the whole CPU.

    Yes, i know that :)) but when they designed IVY they should have designed every chip with the HD 4000. The HD 4000 is still outperformed by Liano iGPU if you remember...
    7
  • On the desktop side yes. Not on the mobile side.
    -2
  • Cant u put an older/previous generation desktop and benchmark against it? I cant or couldnt get how a fast a 2.0GHz Ivy vs a similar Nehelem Desktop CPU vs desktop core 2 duo CPU. Many of us buy notebook to replace desktop for casual use, we would like to know what it can do vs our old desktop.

    Besides get some older AAA games to bench, nobody play BF3 at Ultra books with HD4000. We wanna see what old games we can max out @ full resolution.
    8
  • I think AMD missed a trick with Llano. Instead of throwing four lowly clocked cores at a mobile processor, perhaps two higher clocked cores would've made much more sense. That way, they could possibly sport a higher clock GPU as well within the same TDP.

    Trinity's lower powered, higher clocked cores already look to have partly made up for this, but until the 17W variant comes along, there's no real indication of how it'll measure up to IB ULV. However, we do know that AMD pairs the slower GPUs with the slower CPUs and vice versa, so there's little chance of a, say, 2C/2T/1M CPU with the 7660G GPU.
    -2
  • ... ok... you compare the Llano to Sandy/Ivy bridge in CPU performance, but not in GPU performance? and why not Sandy/Ivy Bridge to Trinity?
    ... and Adobe Photoshop CS 5? not CS 6?
    4
  • DjEaZy... ok... you compare the Llano to Sandy/Ivy bridge in CPU performance, but not in GPU performance? and why not Sandy/Ivy Bridge to Trinity? ... and Adobe Photoshop CS 5? not CS 6?


    Not sure using CS6 would change that chart in any way as to which processor is faster. Also, Llano's GPU isn't faster than Ivy Bridge though Trinity is. Its margin of performance isn't worth the huge processor hit in performance compared to Ivy though. Trinity will make for a good sub $600 ultrabook though.
    0
  • Ivy Bridge: Big jump for mobile, basically nothing for desktops.
    4
  • DjEaZy said:
    ... ok... you compare the Llano to Sandy/Ivy bridge in CPU performance, but not in GPU performance? and why not Sandy/Ivy Bridge to Trinity? ... and Adobe Photoshop CS 5? not CS 6?

    Yeah, an "interesting" choice...

    Regardless, CPU processing power is already overkill for most (if not all) general users in the laptop/ultrabook segment.

    370203 said:
    Also, Llano's GPU isn't faster than Ivy Bridge though Trinity is. Its margin of performance isn't worth the huge processor hit in performance compared to Ivy though. Trinity will make for a good sub $600 ultrabook though.

    Llano's HD6550D universally beats HD4K in games by a significant margin. Even the lower-level GPUs beat HD4K in games. HD4K was a significant improvement over the HD3K, but HD4K is nowhere near the graphical performance of AMDs APUs, Llano or Trinity.

    Whether that matters for ultrabooks, I honestly couldn't say... Honestly, even though Llano and Trinity APUs kill Ivy Bridge in graphical performance, it's not like many people buy ultrabooks for gaming. How GPGPU and things like OpenCL begin to affect general processing tasks will tell with time. Not that it matters much. To reiterate my initial statement above, overall CPU performance in general is so much more overkill than a majority of the mass public would ever use, it's almost pointless to even compare one from the other these days.
    1
  • Quote:
    Gigabyte's U2442N, for example, qualifies as an Ultrabook, but comes equipped with Nvidia's GeForce GT 640M GPU. Faster configurations are going to command higher prices, naturally, but the U2442N might be worth consideration if gaming on a compact notebook is important to you.
    Do you have some information on the U2442N actually existing? It's looks really nice and I'd pick one up myself if the price is right, but as far as anyone can tell, the thing is vaporware.
    0
  • Well, this is my experience with an ivy ultrabook.

    Aspire S3, with the i5 3317U.

    Exact same ports as the reference, average trackpad. Slightly annoying keyboard.

    Ran Sandra on it. Most scores were higher than Dell's XPS13 with an equivalent SB proc, except memory bandwidth due to a single-channel RAM stick, 4GB DDR3-1600 at 11-11-11-28.

    The Acer has a 500GB HDD with a 20GB SSD as a cache.

    Here are the problems i saw:

    1. Trackpad. Miles away from Apple's stuff; same goes for the gesture support.
    2. Annoying keyboard. Good layout, but annoying.
    3. Heat, damn it. I measured the CPU peaking at 76*C at high GPU and CPU load. Vent's at the back, thankfully, unlike the XPS 13. Still gets too hot over long periods.
    4. If the GPU is under load, heat spreads to the CPU too, and vice versa. obvious consequence of ptting both together.
    5. Couldn't run graphics or GPGPU benchmarks on Sandra. Apparently OpenCL driver support is missing on HD4000.
    6. Can't finish an 8GB HD video conversion (arcsoft, quick sync, general H.246, SimHD) on the battery.
    7. Took 20 mins more than the 1.6GHz SB CPU in the XPS 13 to finish the above stated transcode.
    8. Couldn't play Halo CE at 1366x768 on high without severe stuttering and wildly swinging frame rates. they would swing from 40 to 100 fps continuously in an empty room.
    9. The GPU was pegged at 350MHz irrespective of load. Only saw it spike for a tiny (VERY TINY) interval to 1050 MHz. The Dell went there more frequently, there was just something odd here. Had latest drivers, will recheck soon.
    10. Glossy screen :(
    11. Average battery life. 5 hours i guess. Didn't time but never felt like an awful lot.
    12. BLOATWARE. 80+ processes on startup! WTF! McAfee is SOOOOO bloated! heck even interferes with benchmarks.

    But the transcode mentioned took only 1hr 10 mins which is great considering that a Core 2 Quad Q8400 and a 9600GT took 5 hours to do the same.

    Andrew, i don't know if you saw any of these flaws in your ref model, but i truly believe these things (keyboard, screen, trackpad, and other "experience" factors) should be given more emphasis than raw benchmarks. After all, when you buy a laptop in this price bracket, you either get performance or luxury. Clearly, ultrabooks don't fall into the performance category. If you don't get luxury either, isn't it a failing on Intel (or its partners') part?

    Looking to review the new Vaio ultrabooks too (T model an E14 A, i think)!

    Cheers!

    p.s. Why can't anyone except apple use a decent trackpad? Can't understand.


    EDIT:
    There was an updated driver, ends with 8.15.??.2761. Intel's driver update util couldn't id that an updated version was available. Anyway, did a manual install. Halo much more steady now, it's finally hitting 1050 MHz, but then after a period starts jumping b/w that and 350 MHz. GPU usage did hit 100%, spent most of its time above 60%. fps was usually around 70 in an empty room, gunfights make it drop to 40, explosions brought it down to the teens for a moment.

    No QuickSync improvement, still no openCL support. QuickSync activity doesn't seem to trigger the GPU's higher clock, so it sits on 350 MHz.

    EDIT 2: Apparently OpenCL 1.1 should be supported, but it doesn't work with sandra. Unless TH does some more digging, i'm not sure i'll get an answer.
    7
  • ojasWhy can't anyone except apple use a decent trackpad? Can't understand.

    Not a fan of trackpad. I'd use a mouse when available.
    Also, the newer Ultrabooks should have touch capability.

    So don't really care for trackpad.
    -1
  • milkteaNot a fan of trackpad. I'd use a mouse when available.Also, the newer Ultrabooks should have touch capability.So don't really care for trackpad.

    When available. Touch screen ones aren't retailing yet. Even if they were, i'm not going to keep touching the screen for everything, especially not in notebook mode. I'm sure many would agree.
    1
  • I got one of those $350 APU 15-inch laptops -- popped in a 60GB OCZ Agility3 & 2x4GB GSkill (put the 320GB HDD in a cheap Zantac external case). Blows cold air when it blows any air at all. 2 x 1.6GHz Bobcat cores, single channel 1333mhz RAMs, HD 6310 Graphics, UVD3 ... 18w. I'm like a proud poppa.

    17w Trinity is killer -- 25w quad Trinity will be a monster. Get your nose out of Intel's butt.
    -2
  • Seems like a perfect match for the new Microsoft Surface Pro tablets.
    1
  • Seems like a perfect match for the new Microsoft Surface Pro tablets.
    -3