Hacker News Reader

Comment by notreallyserio | Hacker News Reader

No kidding:

https://daringfireball.net/linked/2019/12/17/sgi-workstation...

> The Octane line’s entry-level product, which comes with a 225-MHz R10000 MIPS processor, 128MB of memory, a 4GB hard drive, and a 20-inch monitor, will fall to $17,995 from $19,995.

Really makes the M1 Ultra look affordable.

View on HN

The Indy, which predated the Octane, started much lower ($5k according to Wikipedia, presumably in mid-nineties dollars), but yeah your point very much stands.

The Indy, though, was notoriously underpowered. Very much the “glorified home computer” the GP described, albeit running MIPS.

Still, sure did stand out in the MIT computer labs!

Indys weren't truly that slow. The problem was the base models were memory constrained to the point where IRIX could barely boot. 16MB was not enough, and IRIX 5.x had memory leaks that made it even worse. An Indy with 96MB+ will run IRIX 6.5 pretty well.

That sounds right. I believe most or all developers at the place I worked had either 32 or 64 MB in their machines. At first (~1995) most were probably using IRIX 5.3, but by 96 or 97 I think most if not all had moved to 6.5.

Whatever I had, I don't recall lack of memory ever being a problem. And the GUI was quite snappy.

The GUI was fantastic. Minimal got out of your way as much as possible and used the hardware acceleration to great effect. IRIX 6.5 was rock solid, I used it as my main driver for years before switching to Linux, we also had some windows boxes floating around because we supported a windows binary but that and admin were the only things done on those, everything was either SGI or Linux. I was still using my SGI keyboard two years ago but it finally died.

I seem to remember a very classic rant about how bad Irix was back then but I can't seem to find it.

SGI some places did a great job at giving good deals to computer labs. When I was at university in Oslo, there were rows and rows of Indy's on one side of the biggest undergrad computer lab, and then a bunch of Suns with multiple monochrome Tandberg terminals hooked up on the other.

No big surprise that the Indy side always filled up first, and that "everyone" soon had XEarth and similar running as backgrounds on the Indys... Of course "everyone" loved SGI and were thoroughly unimpressed with Sun after a semester in those labs.

There's a running joke about the Indy that it's the Indigo (its much-more-expensive brother) without the "go".

> Really makes the M1 Ultra look affordable.

The amount of power you can buy today for under $1000 let alone under $10000 is insane compared to back then. The M1 Ultra is not that expensive compared to mid-range workstations or even high-end PCs of previous eras.

Yeah, I don't quite get the way people sometimes reminisce about the hardware costs of the past. We used to have consumer CPUs topping $1000 back when that was some serious money, and big-boy graphics workstations could easily run the tens or hundreds of thousands.

Non-toy computers were only available to the relatively wealthy for nearly a decade. The original Apple II was the equivalent of around $5000, which certainly wasn't a casual purchase for most people.

If you look in back-issues of Byte the prices of early PCs with a usable spec are eye-watering, even before correcting for inflation.

Prices didn't start dropping to more accessible levels until the 90s.

An Irix workstation is like buying a Countach. It's a thing you have because it's the thing that you always wanted.

I used to run an e-mail service with ~2m user accounts '99-'01. Our storage was an IBM ESS "Shark" stocked with 1.5TB of drives and two RS/6000 servers as the storage controllers.

Add on web frontends and mail exchangers, and the entire system was slower and had less aggregate RAM and processing power, less (and slower) disk (well SSD in my laptop) than my current $1500 laptop.

That's just over $31,000 in 2022 dollars. I don't think I can even imagine what kind of modern desktop you could build for that much money.

Could easily get there with a Mac Pro: https://www.apple.com/shop/buy-mac/mac-pro/tower

I'm not even sure you could build a $31,000 desktop computer even if you wanted to without resorting to some ridiculous "expensive for the sake of being expensive" parts. Even quad RTX 3090 Ti's would only set you back $8,000 if you got them at MSRP.

EDIT: Just saw the other comment and I stand corrected.

You can run up costs pretty much arbitrarily with big memory and big storage. 2TB of RAM in a workstation will run you at least $30k if not more (it was $40k last time I checked), and you can go as high as 4TB in current systems. And big storage and NVMe arrays, it's almost a matter of "how much you got?", you can really scale capacity arbitrarily large if you've got the cash (although performance won't increase past a certain point).

This was always the dumb bit with the "apple wants HOW MUCH for a mac pro!?!?" articles about the "$50k mac"... it had $40k of memory in it alone, and the "comparable" systems he was building maxed out at 256GB theoretical and 128GB actual. That's great if it works, using a lower spec will push costs down on both sides, but it's not comparable.

Top-of-the-line 512GB LRDIMM DDR4 will run you about $2,500 before tax if you buy name brand Samsung. I know this because that is what is in both of my dual Xeon workstations. It gets pricey when you go through Dell or HP of course.

> 2TB of RAM in a workstation will run you at least $30k

The trick here is to use a board with 32 dimm sockets -- which requires an oddball formfactor-- but it radically lowers the cost of reaching 2TB.

But your point remains, change your target to 4TB ram (which really isn't an absurd amount of ram) and the astronomical costs come back (unless you go to 96 dimm socket systems, which have their own astronomical costs).

>I'm not even sure you could build a $31,000 desktop

A decked out Mac Pro can reach over $50,000 and it's not even that powerful as your 2x 3090Ti example, but that's the Apple tax for you.

Have you tried to price out a comparative PC? If you even can? Because there aren't many that will take as much RAM as that $50K Mac Pro and when you do find a PC that will all of the sudden you realize there isn't much of an Apple tax at all for the equivalent hardware.

Want to argue that Apple should have more variation in their offerings and price points? Sure - I heartily agree. But blithely tossing out a contextless $50K price tag as being some sort of "tax" is just silly.

>Have you tried to price out a comparative PC? Because there aren't many that will take as much RAM as that $50K Mac Pro and when you do find a PC that will all of the sudden you realize there isn't much of an Apple tax at all for the equivalent hardware

You're deluding yourself with your claim. If you would have done 30 seconds of googling, you would have found out that Dell, HP, Lenovo and Puget Systems, all sell workstations that can beat the top spec Mac Pro or can build one yourself for way cheaper than them but as a business you want the support and no headache so it makes sense to buy prebuilts.

So yeah, there's an Apple tax on it.

https://www.imore.com/53000-pc-competition-apples-53000-mac-...

It’s reasonably easy to build a threadripper workstation with 1TB of RAM (upgrade able to 2TB) and 2 3090’s for $20,000.

LambdaLabs will even sell you a prebuilt around that price with the same specs.

That said, I wouldn’t personally get into a pissing match over x86 Apples. I think their ARM offerings are far more interesting.

Quad RTX A6000s would be $24k and that’s what would go in a “workstation”

Venturing off-topic a bit here, but what exactly makes a "workstation" GPU? What's the difference between an RTX A6000 and an RTX 3090?

ECC RAM, different cooling setup (blower vs side fans), very different thermal characteristics, 24GB or 48GB, more bus width usually, optimized paths for data load & unload, GPU interconnects for direct GPU to GPU communication, shareable vGPUs between VMs, GPU store & halt, h/w support for desktop state, GPU state hand-off to another machine. It isn't just a "more memory" kind of thing.

> GPU interconnects for direct GPU to GPU communication

Note that nvlink is also available on the 3090, which does also have the same bus width as the A6000.

So interestingly to get multiple RTX 3090 cards using NVLINK you really need the blower style, which NVidia recently stopped making in the consumer line, otherwise you are fighting thermal properties and getting in to exotic cooling solutions. And you also get in to placement in slots on the motherboard on three-slot cards. And then you get into the number of PCIe PSU connectors on most supplies. You also get in to weird corner cases of four way NVLINK on 3090's acting as a "single card" but four way NVLINK on RTX A5000 is still independent cards as far as the OS goes. Regarding NVLINK, 3090 cards only support NVLINK (NVL 3.0 IIRC), rather than NVSWITCH. With NVLINK/NVSWITCH, with two RTX A5000 cards I get a total of 48GB of usable RAM for storing the machine learning models, and so on for multiple cards in the RTX A5K & A6K & AX00 line. But with the consumer cards, even though each card has, e.g. 12GB of RAM on them, your deep learning model is also limited to 12GB of RAM, no matter how many 3090 cards you put in. You're also limited to a maximum of four 3090 cards in total, whereas with the NVLINK & NVSWITCH on the professional grade cards, you are limited only by how many cards you can stuff in your data center - technically it is 16 GPUs on a single switch IIRC, things get funky with the fabric once you progress beyond 16 GPUs on a single switch but that's outside of the scope of this conversation. I may also have some details wrong, my brain is a little fuzzy afer dinner.

The A6000 has ECC memory and the 3090 does not. I think that's the chief differentiation between workstations and any other kind of desktop computer. Like a server, they will have ECC everywhere.

In addition to the other stuff people posted you also get to use the certified GPU drivers. Which means they actually tested that the card would work 100% with AutoCAD or whatever

The main difference is you get twice the memory. If you don't need that, there is very little reason to get an A6000.

The RTX is really slow for FP64.

https://zworkstations.com/configurations/3010617/

24x 4.5GHz cores, 96GB memory, 48TB NVMe storage, 2 giant GPUs, etc.

That's wild, although it seems to be server parts in a workstation case. I guess none of Intel's "desktop" chips support a dual/quad CPU configuration though, so that's your only choice. Quad 8 TB NVMe drives is definitely one way to get to $30K of parts pretty quickly.

Neither Intel or AMD support SMP with consumer chips. To go dual-processor with AMD you have to buy EPYC skus, which are several times more expensive than their threadripper core-count equivalents.

FWIW, EPYCs sell on ebay with $/core prices much closer to threadripper prices-- presumably that's closer to what AMD is selling them for to large companies after discounts.

The MSRP on them is ... quite staggering though!

Well dual XEON SP2 CPUs, multiple RTX A5000 GPUs, 30TB of SSD storage, 512GB of RAM and dual BlackMagic quad-input 4K capture cards can get you pretty darn close when it comes to your computer vision work.

You can easily get well past $40k once you start adding some Quadro GPUs, 192gb RAM and a few TBs of PCIe storage into any of the mainstream manufacturers’ workstation products.

Reply on news.ycombinator.com