TR Forums

What is going on with the system guide?

Its not being updated but still on the front page.

Is it dead or are you going to update it?

"It's dead Jim".... From the looks of things, The Tech Report is on Life Support and the only thing that can save it, would be some Buyer who would run this sort of Site and Forum like a Tech site.

I'm not sure that right now is a great time to build, anyway. Lots of retailers are running short on stock these days.

derFunkenstein wrote:

I'm not sure that right now is a great time to build, anyway. Lots of retailers are running short on stock these days.

Thanks for mentioning that as I haven't spec a system out recently... Not that I'm building a system anytime soon.

The fourth quarter of this year is going to be an AWESOME time to build a new PC.

If you have the budget for it, I agree. Sub-$1200 or even sub $1500 builds won't change much, but once you get into Ampere 3070 and Ryzen 7 4000 territory it'll be a great time to build.

All I have to do is survive until fall!

I miss the System Guide.

Ampere is looking like to be Turing on steroids and pushing Samsung's 8mn node to its practical limits. The leaked "3090" reference design show that HSF is a triple-sloter using 120mm fans and it appears that it'll use a single 12-pin PCIe power connector (Nvidia is supplying 8+8 to 12-pin adapter). I wouldn't be surprised that 3090 consumes almost 400W at maximum load.

Custom designs will have THREE 150-watt 8-pin PCIe connectors. The PCIe x16 slot can provide up to 75 watts from the motherboard, so that puts the new Ampere’s power consumption in the range of 376 to 525 watts.

The new micro 12-pin connector is rated for up to 600 watts. It’s significantly more compact than existing PCIe connectors.

JustAnEngineer wrote:

Custom designs will have THREE 150-watt 8-pin PCIe connectors. The PCIe x16 slot can provide up to 75 watts from the motherboard, so that puts the new Ampere’s power consumption in the range of 376 to 525 watts.

The new micro 12-pin connector is rated for up to 600 watts. It’s significantly more compact than existing PCIe connectors.

I wouldn't call 12-pin connector "micro". It is just a scale-up PCIe connector with extra wiring for power. I suspect most potential buyers will probably get 3070 and 3080 SKUs which will probably fall somewhere between exceeding 2080 Super and exceeding 2080Ti respectively while 3090 goes out beyond no matter the cost. Ampere appears that have little or no under the hood improvements is relying much more on brute force to outpace its predecessor. It doesn't help it stuck on a inferior process node since Nvidia pissed off TSMC when they tried to haggle hard.

Krogoth wrote:

JustAnEngineer wrote:

The new micro 12-pin connector is rated for up to 600 watts. It’s significantly more compact than existing PCIe connectors.

I wouldn't call 12-pin connector "micro". It is just a scale-up PCIe connector with extra wiring for power.

It has more pins, but they're smaller. One 600-watt micro 12-pin connector is about the same length and only 2/3 as wide as a single 150-watt 8-pin PCIe connector.
https://www.techpowerup.com/271301/nvid ... e-its-tiny

Krogoth wrote:

I wouldn't call 12-pin connector "micro".

Well, that's what Molex/Seasonic are apparently calling it, so assuming the TPU story is accurate, you would be wrong...

JustAnEngineer wrote:

Custom designs will have THREE 150-watt 8-pin PCIe connectors. The PCIe x16 slot can provide up to 75 watts from the motherboard, so that puts the new Ampere’s power consumption in the range of 376 to 525 watts.

The new micro 12-pin connector is rated for up to 600 watts. It’s significantly more compact than existing PCIe connectors.

Good Grief are they trying to find justification for gamers to buy 1,000+ watt PSUs? Of course most gamers are not going spend their hard earned money silly ass priced very high end dGPUs that also need such PSUs with such a high wattage anyway. But still...

If I'm spending that much on just a couple of computer parts, then they better be for something that makes me a lot more money then what they cost me. And nothing nothing silly like Mining for the latest Coin Fad either.

They're recommending an 850-watt power supply.

JustAnEngineer wrote:

They're recommending an 850-watt power supply.

That is still a high wattage PSU. As far as I'm concern, anything above a 550w Gold Standard PSU is way too much to be buying.

Remember that Intel's current 14 nm Comet Lake CPUs are also extremely power-hungry, pulling more than double their "TDP" rating, so the 850-watt PSU isn't all for the Ampere GPU.

If you can afford what will be a graphics card that costs $1K you can shell out a little more for a PSU.

JustAnEngineer wrote:

Remember that Intel's current 14 nm Comet Lake CPUs are also extremely power-hungry, pulling more than double their "TDP" rating, so the 850-watt PSU isn't all for the Ampere GPU.

I really haven't been keeping with the latest and Greatest, but Intel is still at 14nm? AMD is already using 7nm for both it's CPUs and dGPU. Doesn't Apple has samples of it's ARM SoC made with 7nm tech? Meanwhile I'm not sure about Nvidia.

tfp wrote:

If you can afford what will be a graphics card that costs $1K you can shell out a little more for a PSU.

While true, I do believe that spending that much money is stupid even if you can easily afford the money. Unless you making a great of it from the hardware.

Next up as qualified PSUs? Industrial arc welders.

Captain Ned wrote:

Next up as qualified PSUs? Industrial arc welders.

How many watts do those things use anyway?

They output 800W at 1 ohm impedance, if you're talking about a pair of Mark Levinson stereo amps used as a welder. This was the beastie one used to drive Apogee Full Range or Scintilla true ribbon speakers.

https://www.stereo.net.au/forums/topic/ ... nt=1727042

Captain Ned wrote:

They output 800W at 1 ohm impedance, if you're talking about a pair of Mark Levinson stereo amps used as a welder. This was the beastie one used to drive Apogee Full Range or Scintilla true ribbon speakers.

https://www.stereo.net.au/forums/topic/ ... nt=1727042

Wait I'm confused, are We talking about tools for welding two pieces of metal together, or something that is used to power Speakers?

Well, the ML-20 can do both.

Sorry. I was a subscriber to Audio when that article first ran and ever since I've been "is it an amplifier or an arc welder".

whm1974 wrote:

I really haven't been keeping with the latest and Greatest, but Intel is still at 14nm? AMD is already using 7nm for both it's CPUs and dGPU. Doesn't Apple has samples of it's ARM SoC made with 7nm tech? Meanwhile I'm not sure about Nvidia.

It's a bit of an apples/oranges comparison since Intel and TSMC define their process nodes slightly differently, but yeah. Intel has fallen behind in the process race. All the experience making smartphone SOCs has catapulted TSMC to the bleeding edge, and AMD is leveraging TSMC's process to the hilt. GloFo has thrown in the towel on bleeding edge CPU manufacturing, and is relegated to making the Ryzen I/O die.

just brew it! wrote:

whm1974 wrote:

I really haven't been keeping with the latest and Greatest, but Intel is still at 14nm? AMD is already using 7nm for both it's CPUs and dGPU. Doesn't Apple has samples of it's ARM SoC made with 7nm tech? Meanwhile I'm not sure about Nvidia.

It's a bit of an apples/oranges comparison since Intel and TSMC define their process nodes slightly differently, but yeah. Intel has fallen behind in the process race. All the experience making smartphone SOCs has catapulted TSMC to the bleeding edge, and AMD is leveraging TSMC's process to the hilt. GloFo has thrown in the towel on bleeding edge CPU manufacturing, and is relegated to making the Ryzen I/O die.

Now that TSMC is on it's way to 5nm if not upgrading Fabs, and even researching 3nm, Intel is getting really getting behind. Given Intel's history of investing Profits in improving Node Process this highly unusual for them.

All of the world's leading semiconductor foundries buy their lithography equipment from ASML. A new Twinscan NXE 3400B EUV lithography machine costs $120,000,000. TSMC's leadership has a vision of how to grow their business and make it profitable. They have invested billions in the past several years to develop and commercialize the new technology and they're reaping the benefits. They now own half of the EUV lithography systems that have been delivered and they are producing 60% of the world's EUV imaged wafers.

Intel, meanwhile, has recently announced another $2,400,000,000 of stock buybacks to temporarily prop up their sagging stock price. The cash spent on this buyback could have purchased 20 new EUV lithography systems. Instead, Intel proceeds to bring out another CPU "generation" using ye olde DUV at 14 nm.

We've got $2,400,000,000 in cash sitting here. Should we...
1) Buy a complementary business to grow our company, thereby increasing profit now and in future years?
2) Upgrade our manufacturing capability to increase sales volumes and revenue, thereby increasing profit now and in future years?
3) Upgrade our infrastructure to make make us more efficient and reduce costs, thereby increasing profit now and in future years?
4) Sit on the cash for now, maintaining financial security and flexibility until a good opportunity comes along in the future?
5) Buy back stock to make the company poorer and smaller, but achieve short-term increases in share prices to make our fat stock options more valuable?

This shows the difference between company leadership that has a vision and believes in their business and leadership that's just interested in the short-term appearance of success.

The question is, does Intel have enough cash flow to do both the stock by back and buy enough lithography system? They also pay a quarterly dividend. They are a big enough company, have enough cash, and cash flow to do more than 1 thing.

tfp wrote:

The question is, does Intel have enough cash flow to do both the stock by back and buy enough lithography system? They also pay a quarterly dividend. They are a big enough company, have enough cash, and cash flow to do more than 1 thing.

Whether it is an issue with equipment or with having sufficient process engineering staff (or both), it seems pretty clear that they haven't been doing what it takes lately.