VivienM wrote on 2024-08-31, 14:34:

It could be about processor instructions too - they use newer compiler that uses newer instructions, and oops, that sets a floor […]
Show full quote

Namrok wrote on 2024-08-31, 14:13:

That aside, sometimes I see the system requirements on games like StarCraft HD and I weep. The original game ran on a Pentium 90. How on earth does higher resolution sprites require, lets see, a Geforce 6800 and an Athlon 64 X2? I can only assume the engine or middlewear in the remaster experienced profound bloat. I recall an anecdote around the development of Daikatana where an artist they hired who'd never done games before handed in some preposterously sized bitmap, thousands of pixels wide and tall, for an icon that would be 32x32 pixels on screen. And that just feels emblematic of all software now.

It could be about processor instructions too - they use newer compiler that uses newer instructions, and oops, that sets a floor on what processors can run the thing. Athlon 64 X2/Pentium D could be a rough shortcut for "need SSE3" and they don't want to put lots of asterisks about how this core Athlon 64 and that core Pentium 4 is fine but the other core is not.

But also... is an Athlon 64 X2 that unreasonable a requirement? StarCraft Remastered came out in 2017; the Athlon X2's glory days were in 2005, 12 years earlier. The recommended system for StarCraft Remastered is an E6600/8800GT; I had one of those in 2007. Meanwhile, original StarCraft came out in March 1998, i.e. when that Pentium 90 would have been just four years old.

One other thing I would be very curious about - actual performance. Back in the 1990s, minimum requirements for software were... ridiculous. Recommended requirements were barely passable. So my guess is that original StarCraft on that P90 was barely playable. What is StarCraft Remastered like on an X2 3800+ with a 6800 and 2GB of RAM running Win7 32-bit? Is it liveable?

I mean, you aren't wrong in that sense. Having the minimum system requirement set to an Athlon 64 X2 in 2017 when it came out is basically saying "This will run on any computer you likely still have." I'm more commenting on the fact that, for example, my Ryzen 5800X3D has 100's, if now 1000's of times the processing power of my families first DX2, or my own first Pentium 120. But I am not getting 100's or 1000's of times the experience out of it. 10x the experience out of it, tops. And in the last 10 years, it's almost been a regression, where the more powerful home computers get, bafflingly, the software experience is actually moving backwards!

It's sad to hear that they shut down. I hope the website gets archived because it was a good source of information on old hardware. I haven't read any of their newer reviews because I haven't been interested in modern hardware since Sandy Bridge was such a huge disappointment (but apparently only for me).

On the (off-topic) subject of ARM PCs, I don't think Microsoft and Qualcomm are the right companies to make it work. Microsoft is totally incompetent, and Qualcomm is just trying to hack cell phone SoCs (with loads of corporate lock-in bullshit like proprietary charging protocols over USB-C or locking board designers into specific PMICs that aren't quite adequate for the application) into laptops because that's all they know.

For what it's worth, Gamers Nexus actually began writing reviews some time ago.

Namrok wrote on 2024-08-31, 15:40:

I'm more commenting on the fact that, for example, my Ryzen 5800X3D has 100's, if now 1000's of times the processing power of my families first DX2, or my own first Pentium 120. But I am not getting 100's or 1000's of times the experience out of it. 10x the experience out of it, tops. And in the last 10 years, it's almost been a regression, where the more powerful home computers get, bafflingly, the software experience is actually moving backwards!

Oh, I know. If you want to see this in a more extreme way, go and play with a vintage PPC Mac, like a G4 MDD, with plenty of RAM. The nice thing about the PPC Mac platform is that it died an abrupt death around 2009. It never had Chrome, Chromium, etc.

So it's frozen in time in a way that an XP (or Vista, which is the OS that most matches with Leopard) x86 machine wouldn't be. The XP machine can run much newer stuff, run it more slowly, etc, there isn't this abrupt cut-off.

The G4 Mac, even with a hard drive, is... fast. Very fast. Many storage-intensive tasks (e.g. OS patching) are actually faster than the equivalent patching on modern systems on stupid fast SSDs. Or at least, it's fast until you attempt to run one of the few retro web browsers available for it, in which case you discover it is utterly useless.

I continue to believe that every innovation in desktop computing in the past 10-15 years has gone to feeding the monstrosity known as the modern web, web technologies, and all of its derivatives. You now have a world where almost every program has an embedded Chromium just so it can do 'modern authentication'. And sometimes Moore's law has just... not been enough... to feed the modern web.

But fundamentally, I agree with you. The move to real operating systems (i.e. not 98SE) was significant. But am I getting that much more of my modern systems that I didn't get from my PIII 700 with 640 megs of RAM running Win2000? Probably... not. My Windows desktop has 64 gigs of RAM, the iMac I am typing this on has 128 gigs of RAM, I don't know how much more CPU performance they have than the PIII 700, but I certainly am not doing 100X more.

lti wrote on 2024-08-31, 15:49:

It's sad to hear that they shut down. I hope the website gets archived because it was a good source of information on old hardware. I haven't read any of their newer reviews because I haven't been interested in modern hardware since Sandy Bridge was such a huge disappointment (but apparently only for me).

And that, right there, is why they shut down.

If you think about it, back in the 1995-2010 period, you were hoping for a 5 year lifecycle on your desktop computer, and that was a stretch. So you go from your late PIII or Athlon XP or Willamette/early Northwood in 2001ish to your Conroe E6600/Q6600 in 2006-7, and you're expecting a similar improvement in 2011-2ish, and... that improvement is just not there. Your E6600/Q6600 isn't struggling the way your late PIII was in 2006, the benchmarks on the Sandy Bridge are... meh, and your interest level just starts dropping.

I still read AnandTech's articles (the wonders of RSS...), but if it didn't show up in my RSS feed, I probably wouldn't go seek them out.

(Also, I tend to agree that Sandy Bridge was somewhat of a disappointment at the time. It's only when everything that came after Sandy/Ivy Bridge turned out to be even more of a disappointment that Sandy/Ivy Bridge retroactively became this legendary platform. It's funny, in 2010 I reorganized some hardware around, made a Q8300 primarily out of existing parts my main desktop box thinking that that would be temporary due to how weird the Nehalem lineup was and it'd be replaced by the next thing after the Nehalem mess, and that Q8300 ended up lasting me until 2017 when I got an i7-7700)

VivienM wrote on 2024-08-31, 16:17:

lti wrote on 2024-08-31, 15:49:

I haven't been interested in modern hardware since Sandy Bridge was such a huge disappointment (but apparently only for me).

(Also, I tend to agree that Sandy Bridge was somewhat of a disappointment at the time. It's only when everything that came after Sandy/Ivy Bridge turned out to be even more of a disappointment that Sandy/Ivy Bridge retroactively became this legendary platform. It's funny, in 2010 I reorganized some hardware around, made a Q8300 primarily out of existing parts my main desktop box thinking that that would be temporary due to how weird the Nehalem lineup was and it'd be replaced by the next thing after the Nehalem mess, and that Q8300 ended up lasting me until 2017 when I got an i7-7700)

Not sure what you guys are remembering, the 2600K was highly lauded when it arrived and practically everyone wanted one.

I was a regular reader from 1998 (waiting eagerly for any new hardware related stuff) to around 2005 and then occasionally to around 2010. After that maybe a few times a year. So I guess I won't miss it now. But it would be great if the archive was still available for reading, especially the early years.

jmarsh wrote on 2024-08-31, 16:29:

VivienM wrote on 2024-08-31, 16:17:

lti wrote on 2024-08-31, 15:49:

I haven't been interested in modern hardware since Sandy Bridge was such a huge disappointment (but apparently only for me).

(Also, I tend to agree that Sandy Bridge was somewhat of a disappointment at the time. It's only when everything that came after Sandy/Ivy Bridge turned out to be even more of a disappointment that Sandy/Ivy Bridge retroactively became this legendary platform. It's funny, in 2010 I reorganized some hardware around, made a Q8300 primarily out of existing parts my main desktop box thinking that that would be temporary due to how weird the Nehalem lineup was and it'd be replaced by the next thing after the Nehalem mess, and that Q8300 ended up lasting me until 2017 when I got an i7-7700)

Not sure what you guys are remembering, the 2600K was highly lauded when it arrived and practically everyone wanted one.

So, in honour of this thread's original focus, I went to Anand's review of the Sandy Bridges (https://www.anandtech.com/show/4083/the-sandy … -i3-2100-tested ). Checked a few of his charts. He is showing about 40% higher performance than a Q6600, which was the 'go to' chip for many enthusiasts (I missed out on the Q6600 because I was so starved for a new system I preordered an E6600, long before the Q6600 was announced and became affordable).

What was the go-to chip for most enthusiasts before the Q6600? The Athlon X2 3800+? The 3.06GHz Northwood? Without checking Anand's reviews, I'm pretty sure the Q6600, or even the E6600, was more than 40% faster than those.

The big price drop that made the Q6600 a legend was in July 2007. So in 3.5 years between that and the Sandy Bridge launch, you get... 40% performance increase, assuming stock clocks on both? And in those 3.5 years, the software side stagnates - it's not like Windows hardware requirements went up between that time, if anything 7 is a little more optimized than Vista...

The other thing I would note - look at the rest of the motherboard. Going from DDR1 to DDR2 upped the practical max RAM from 2GB to 8GB. SATA was added. PCI-E replaced AGP. USB 2. There were lots of platform improvements prior to Sandy Bridge, but... compared to whatever board you were running your Q6600 on, Sandy Bridge added nothing other than a buggy higher-speed SATA mode. Didn't even include on-chipset USB 3; that would come in Ivy Bridge. DDR3 would eventually mean higher RAM capacity but I don't think that mattered in 2011.

Man, this seriously bites! I remember very enthusiastically purchasing an Athlon 64 3200+ and a Ryzen 9 5950X just days after reading their respective reviews on AnandTech. And a GeForce 6800GT, maybe? Although, that one may have been the Tech Report's doing -- another awesome site we lost a few years ago.

AT really was one of the greats. Just loved their HTPC reviews and tips back in the day; iirc Anand and Ganesh were real home theater nuts!

VivienM wrote on 2024-08-31, 17:03:

So, in honour of this thread's original focus, I went to Anand's review of the Sandy Bridges (https://www.anandtech.com/show/4083/the-sandy … -i3-2100-tested ). Checked a few of his charts. He is showing about 40% higher performance than a Q6600, which was the 'go to' chip for many enthusiasts (I missed out on the Q6600 because I was so starved for a new system I preordered an E6600, long before the Q6600 was announced and became affordable).

What was the go-to chip for most enthusiasts before the Q6600? The Athlon X2 3800+? The 3.06GHz Northwood? Without checking Anand's reviews, I'm pretty sure the Q6600, or even the E6600, was more than 40% faster than those.

Maybe because your "40%" is a backwards calculation? Looking at those tests, in most of them the 2600K recorded times that were twice as fast as the Q6600, i.e. it was 200% as fast, with a relative increase of 100%.

jmarsh wrote on 2024-08-31, 18:03:

VivienM wrote on 2024-08-31, 17:03:

So, in honour of this thread's original focus, I went to Anand's review of the Sandy Bridges (https://www.anandtech.com/show/4083/the-sandy … -i3-2100-tested ). Checked a few of his charts. He is showing about 40% higher performance than a Q6600, which was the 'go to' chip for many enthusiasts (I missed out on the Q6600 because I was so starved for a new system I preordered an E6600, long before the Q6600 was announced and became affordable).

What was the go-to chip for most enthusiasts before the Q6600? The Athlon X2 3800+? The 3.06GHz Northwood? Without checking Anand's reviews, I'm pretty sure the Q6600, or even the E6600, was more than 40% faster than those.

Maybe because your "40%" is a backwards calculation? Looking at those tests, in most of them the 2600K recorded times that were twice as fast as the Q6600, i.e. it was 200% as fast, with a relative increase of 100%.

I was looking at the frame rates on the gaming benchmarks page; if the Q6600 gets 100 and the sandy bridge gets 140, isn't that 40% faster?

VivienM wrote on 2024-08-31, 18:27:

I was looking at the frame rates on the gaming benchmarks page; if the Q6600 gets 100 and the sandy bridge gets 140, isn't that 40% faster?

Direct text from the game benchmarks page:

There's simply no better gaming CPU on the market today than Sandy Bridge. The Core i5 2500K and 2600K top the charts regardless of game. If you're building a new gaming box, you'll want a SNB in it.

That sure sounds like Sandy Bridge was a step up to me...
Regardless, game benchmarks aren't a great test for comparing CPUs because of the amount of work being offloaded to the GPU. As is apparent in the latter tests for the (mostly) 2D games like Starcraft2 and Civ V, where the Q6600 doesn't even make it into the results table.

jmarsh wrote on 2024-08-31, 18:48:

Direct text from the game benchmarks page: […]
Show full quote

VivienM wrote on 2024-08-31, 18:27:

I was looking at the frame rates on the gaming benchmarks page; if the Q6600 gets 100 and the sandy bridge gets 140, isn't that 40% faster?

Direct text from the game benchmarks page:

There's simply no better gaming CPU on the market today than Sandy Bridge. The Core i5 2500K and 2600K top the charts regardless of game. If you're building a new gaming box, you'll want a SNB in it.

That sure sounds like Sandy Bridge was a step up to me...
Regardless, game benchmarks aren't a great test for comparing CPUs because of the amount of work being offloaded to the GPU. As is apparent in the latter tests for the (mostly) 2D games like Starcraft2 and Civ V, where the Q6600 doesn't even make it into the results table.

I don't think anyone is saying Sandy Bridge wasn't a step up (or that it was a bad processor), we're saying it was perceived as a disappointment at the time, especially compared to the pace of innovation in the previous decade?

If you take, say, your $300-in-mid-2007 Q6600, that's 4 Conroe cores. One Conroe core probably has... at least double... the performance of what most people were upgrading from (closer to 40% more if you were coming from 3800+ AMDs, but many people were coming from much older - the 3800+ X2 was only two years old at the time of the big Q6600 price drop). And other than people who had 3800+ X2 Athlons on 939 (i.e. people upgrading every two years), pretty much everybody else was coming from single core, so... basically, assuming a multi-core workload, you're looking at probably a potential ~8X performance increase.

Is Sandy Bridge 6-8X faster than what most people had when they pulled up Anand's article in January 2011? At most, you're getting double the core count; if you were already on Q6600 or other quad cores, you're gaining no cores. So you're left... with just the boost in single core performance. A reasonable boost in single core performance, sure, but...

And look at Anand's wording carefully - he's saying that "if you're building a new gaming box" in 2011, you'll want a Sandy Bridge. Meanwhile, look at his words in the conclusion to his Conroe review: "However, if all goes well, although Vista may be delayed until 2007, Intel's Core 2 processors will give you a very good reason to upgrade this year."

In other words, Conroe is so good that you should go out and upgrade your desktop, while Sandy Bridge is what you should go with if you've already decided to upgrade your desktop.

(And there have been other reviews like that, e.g. I upgraded my video card from a 7900GT to an 8800GT less than 18 months after getting the 7900GT. Go and read AnandTech's review of the 8800GT - https://www.anandtech.com/show/2365/14 - this is the kind of review that causes starving students to dig up their credit cards and order an 8800GT. And the benchmarks are showing close to 3X higher frame rates over the 7950GT, i.e. last year's product, at higher resolutions. It's funny - I had completely forgotten why I upgraded that video card so quickly, and reading AnandTech's review, now I remember. Meanwhile, the Sandy Bridge review wouldn't cause anybody who wasn't planning to build a new box the day before to look for their credit card and order one. )

VivienM wrote on 2024-08-31, 02:33:

Cool peripherals just exist less and less - there are no nice/5.1 computer channels anymore, cool sound cards have been dead, Microsoft (my favourite vendor) stopped making nice keyboard/mice, no storage peripherals (still looking for a good way to back up my NAS), etc.

The last innovative peripherals (and games) to hit was when VR made a resurgence back in 2016. Going from just watching games on a screen to being inside the game is pretty transformative of the gaming experience. Certainly it was the last time I remember being legitimately excited for new tech.

Unfortunately the PCVR ecosystem never got much traction and it's been limping along since then.

Shponglefan wrote on 2024-08-31, 19:49:

Unfortunately the PCVR ecosystem never got much traction and it's been limping along since then.

Probably because people aren't too keen on wearing a heavy, uncomfortable thing on their head while barely avoiding getting entangled in its cables. Then there's the motion sickness and headaches. Fun.

Wake me up when they make a proper Holodeck. Until then, VR can miss me entirely.

Wouldn't be a holodek - but what if you were surrounded by screens on all sides that would change as you "walk" through? There is an idea

Joseph_Joestar wrote on 2024-08-31, 19:57:

Probably because people aren't too keen on wearing a heavy, uncomfortable thing on their head while barely avoiding getting entangled in its cables. Then there's the motion sickness and headaches. Fun.

Those generally weren't an issue for me. The only ergonomic issue I found was heat. Since VR gaming is more physically active, I always needed a cool ambient temperature to remain comfortable. This generally meant I couldn't play in the VR summer when it would be too warm.

The only time I remember getting nauseated was when I tried to play Metroid Prime in VR. The constant switching between first-person and 3rd-person was highly disorienting.

But games that are designed properly for VR tend to not have motion sickness issues.

ncmark wrote on 2024-08-31, 20:00:

Wouldn't be a holodek - but what if you were surrounded by screens on all sides that would change as you "walk" through? There is an idea

It wouldn't really be the same without stereoscopic imaging. The sense of depth and scale is one of the big things that sets VR apart from just looking a regular screen, no matter how big that screen is.

I guess I missed my 10 year mark last year.

This one is kind of a bummer... I have fond memories of reading Anandtech and researching my first PC build with a Computer Shopper magazine. Ended up getting an M-Tech R534 to run a P5-150 at a then exotic 75MHz FSB.

This grassroots enthusiast journalism of the era, along with Tom's/HardOCP, were seemingly the only ones testing and reporting on Intel's issues with the 1.13GHz P3 that eventually led to a recall.