This forum is in permanent archive mode. Our new active community can be found here.

Rym's New Computer

245

Comments

  • edited January 2010
    Here's a chart of various cards' performance at high settings. I find that performance under heavy stress is also a good indicator of how long the card will last before you need an upgrade. If you just want a card that gets 60+ fps on current games at 1680x1050, then all of the above cards will work well, though.

    On that chart, the lowest group (HD 4850 and GTS 250 = 9800 GTX+) both get 103. The HD 5750 gets 132, so it's around 30% more powerful for 30% more cost. In turn, the HD 5770 gets 152, so it's around 50% more powerful for 50% more cost. On this basis, all of these cards are cost-effective solutions - all 6 of the cards I listed above are solid choices, though the three I picked out are slightly better than the others.

    Personally, I would probably recommend the HD 5750. It offers a new generation of technology at a low price point. Before purchasing I'd take a good look through the available data on its behaviour under Linux, particularly on Phoronix.
    Post edited by lackofcheese on
  • Here's achartof various cards' performance at high settings.
    Thank you for this chart, allow me to use it to point out the hilarity.

    First, look at the title of the chart:

    Sum of FPS Benchmarks 1920x1200 with anti aliasing, 8AA (High Quality)

    Ok, so who is running at 1920x1200 with full anti-aliasing and such? Those are crazy crazy high settings. I run at 1680x1050 with 2x or 4x AA at MOST. I turn all the other options all the way up.

    On this chart, my card, the 8800GT scores a 93.10. That means that using crazy high settings, that are higher than most people will ever use, it runs an fps at over 93 frames per second. You realize that pretty much every monitor only displays 75 frames per second. This means that a cheap piece of crap card that is multiple years old can achieve full vertical sync on a modern game with the settings cranked way way way the fuck up.

    That means that the card I have is about 30% overkill, even today. If my cheap old card is 30% overkill, the cards you people want to buy are super super super overkill.
  • the cards you people want to buy are super super super overkill.
    You realize you're arguing over, at most, $50 on a $1500+ build.
  • edited January 2010
    On this chart, my card, the 8800GT scores a 93.10. That means that using crazy high settings, that are higher than most people will ever use, it runs an fps at over 93 frames per second. You realize that pretty much every monitor only displays 75 frames per second. This means that a cheap piece of crap card that is multiple years old can achieve full vertical sync on a modern game with the settings cranked way way way the fuck up.
    Did you perhaps read the part of your bolded text that says "Sum of FPS Benchmarks"? It's a sum of benchmarks from at least 3 different games. Consequently, your 8800GT is getting less than 30fps in most of these games.

    Also, if you read my post, you would've noticed that the reason I posted the max-settings benchmark was to demonstrate how these cards perform under stress, which is quite relevant to how the card will perform, say, a year or two down the line.

    If you want a more directly relevant example, rather than an attempt to measure overall power of the cards, here's the benchmark for Far Cry 2 (1680x1050, 4AA, 8AF, Very High Quality), an example of a relatively demanding game by today's standards.
    Your 8800GT (presumably 512MB) manages 31.9 FPS, the HD 4850 gets 36.4 FPS, the HD 5750 gets 43.2FPS, and the HD 5770 gets 49.4FPS. Not one of them gets 60FPS. In fact, to get 60FPS+ on this game at these settings for a reasonable price, you're pretty much forced to go with a multiple-card configuration.

    One other point that you're completely missing is that benchmarks like this will typically look at the average framerate over some period of time. Even though you may get an average of, say, 60 fps, that doesn't necessarily mean the game will run near 60fps the whole time. You may still have times when the framerate drops to 30fps or lower. Minimum framerate is a significantly more important benchmark than average framerate.
    Post edited by lackofcheese on
  • Did you perhaps read the part of your bolded text that says "Sumof FPS Benchmarks"? It's a sum of benchmarks from at least 3 different games. Consequently, your 8800GT is getting less than 30fps in most of these games.
    Whaaaaat. I'll have to benchmark, because I am not getting FPS that low. Probably just because the settings are so crazy high in those benchmarks.
  • I think I'm gonna pull the trigger on a 5770 myself. Twice the performance of my 3870, the same wattage requirement, and nice, short PCB.
  • edited January 2010
    Did you perhaps read the part of your bolded text that says "Sumof FPS Benchmarks"? It's a sum of benchmarks from at least 3 different games. Consequently, your 8800GT is getting less than 30fps in most of these games.
    Whaaaaat. I'll have to benchmark, because I am not getting FPS that low. Probably just because the settings are so crazy high in those benchmarks.
    Yeah, the 30% extra pixels for 1920x1200 makes a difference, but the antialiasing in particular is rather demanding indeed.

    In any case, the later benchmark I posted for Far Cry 2 (slightly less demanding than the Crysis engine, according to Wikipedia) is of note. Keep in mind that the minimum framerate for a game will typically be around half of the average achieved - certain scenes in Far Cry 2 would probably drive your card down to 16FPS or so, with these settings.
    I think I'm gonna pull the trigger on a5770myself. Twice the performance of my 3870, the same wattage requirement, and nice, short PCB.
    A solid card. I hope I've been helpful.
    Post edited by lackofcheese on
  • I've done some additional research since I'm buying a computer soon myself. Consequently, I've discovered that the Core i7-920 is no longer a very good choice. The Core i7-860 is faster and slightly less expensive, while the Core i5-750 is only $200 and it's basically just as fast as the i7-920.

    Also, as I said before, you're needlessly spending on the RAM - get the cheapest 3x2GB kit you can (currently at $135).

    Are you going to update your wishlist, Rym?
  • I noticed you are buying the 32-bit version of Windows 7. If I'm not mistaken, can't 32-bit windows only handle a little under 4 gigs, meaning you would have over 2 gigs going to waste?
  • I noticed you are buying the 32-bit version of Windows 7. If I'm not mistaken, can't 32-bit windows only handle a little under 4 gigs, meaning you would have over 2 gigs going to waste?
    I just threw that in. I'm getting 64-bit. ;^)
  • Everything is decided except the video card (wishlist updated).

    I'm choosing between the following two video cards:

    EVGA 896-P3-1255-AR GeForce GTX 260 Core 216 896MB 448-bit GDDR3 PCI Express 2.0 x16 HDCP Ready SLI Supported Video Card
    EVGA 01G-P3-1158-TR GeForce GTS 250 1GB 256-bit DDR3 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card

    The price difference is only $50, and I haven't examined benchmarks yet. This seems to be where the price plateau is before the utter bullshit cards that aren't worth the money.

    I plan to purchase the PC tomorrow. ^_^
  • edited January 2010
    I wouldn't recommend spending that much on a 1GB GTS 250 when you can get a 512MB one for $110. If it has to be one of the exact two cards (brand, price and all) you put forth, I'd go with the GTX 260, but there are GTX 260s that are $30 cheaper.

    Here's the cards I would consider:
    $100 - HD 4850 512MB
    $110 - HD 4850 1GB
    $110 - GTS 250 512MB
    $135 - HD 5750 1GB
    $163 - HD 5770 1GB
    $185 - GTX 260 896MB
    Post edited by lackofcheese on
  • I like my GTX 260, which is where I found the price / performance sweet spot.
  • Well, I would say that the sweet spot occupies the whole $100-$200 range. Unfortunately, the GTX 260 is not priced aggressively right now, unlike the new HD 5750 and 5770.
  • there are GTX 260s that are $30 cheaper.
    And $30 louder. These are quiet.
  • there are GTX 260s that are $30 cheaper.
    And $30 louder. These are quiet.
    Based on what?
  • I like my GTX 260, which is where I found the price / performance sweet spot.
    Same here. It's a great card.
    And $30 louder. These are quiet.
    Solution: Listen to very loud metal until you have permanent moderate hearing loss in both ears!
  • How about this card for quietness?
  • Alright, guys, I need a "tiny god," like Rym said. $500-$700. Can you help me out?
  • edited January 2010
    Well, here's a list I put together for myself for $950:
    lackofcheese's new PC
    I think I did a very good job getting a cheap but rock solid build.
    Post edited by lackofcheese on
  • edited January 2010
    Well, here's a list I put together for myself for $950:
    lackofcheese's new PC
    I think I did a very good job getting a cheap but rock solid build.
    That's a nice build. I'm probably getting something like that once I'm better settled in. I looked up some benchmarks for the GTS 250 and GTX 260.
    fo3PhotobucketPhotobucket
    On this last chart the website warned that the reason for the really low fps is that it's done on a very specific scene where there is a nuclear explosion. Here is the link to the article.
    Post edited by sucrilhos on
  • Rym, I say 260 without even a moments hesitation.
  • Rym, I say 260 without even a moments hesitation.
    I agree. The average framerates are substantively different enough to warrant the small difference in price. I see no reason in the world, however, to consider anything more powerful or expensive.

    Also, for Scott's benefit, I'll note that it is universally agreed that the dual-slot video cards are much quieter than the single-slot ones in most cases.
  • edited January 2010
    I agree. The average framerates are substantively different enough to warrant the small difference in price. I see no reason in the world, however, to consider anything more powerful or expensive.
    It depends on the resolution, though.
    Also, for Scott's benefit, I'll note that it is universally agreed that the dual-slot video cards are much quieter than the single-slot ones in most cases.
    Indeed. It's pretty much self-evident, really.
    Post edited by lackofcheese on
  • I see no reason in the world, however, to consider anything more powerful or expensive.
    The beauty of personal choice.

    My 5770 arrives today. Excited. ^_^
  • Hmm... I think it's time to get a new mouse. Been using the same old logitech for a decade now...
  • edited January 2010
    Hmm... I think it's time to get a new mouse. Been using the same old logitech for a decade now...
    You should get a censor and put it on the belly of a real mouse.
    Post edited by Kate Monster on
  • You should get a censor and put it on the belly of a real mouse.
    Heh, talk about confusing misspellings...
    Hmm... I think it's time to get a new mouse. Been using the same old logitech for a decade now...
    What are you thinking of? I love my 2-year-old MX Revolution, and I've heard the updated version is even better.
  • You should get a censor and put it on the belly of a real mouse.
    That you be the most ineffectual input device ever. Your pointer would go whoopwhoopwhoop all over the screen as the mouse dashes around.
  • You should get a censor and put it on the belly of a real mouse.
    That you be the most ineffectual input device ever. Your pointer would go whoopwhoopwhoop all over the screen as the mouse dashes around.
    Not if you break its legs.
Sign In or Register to comment.