AusWolfFirst thoughts: Dual fan design and 6-pin power? Isn't that excessive for a card like this?

Second thoughts after looking at the heatsink: Oh... It'll probably be noisy as hell anyway.

Overkill, like lipstick on a pig.

Alan SmitheeThe big question is the NVENC/NVDEC feature set. We sorely need a low-end card with decent memory bandwidth for non-gaming (business & entry level content creation, e.g. training session re-edits and basic YouTube). The 1050 had a full encode/decode feature set for its time, but the 1030 did not. The "1630" name thus worries me. Is this really a replacement for the 1050 or the 1030?

I don't know why nvidia didn't choose to call it something along the lines - GT 4005 or something, why do they keep using the GTX 1*** series naming - 6-year-old naming?

Dr. DroUnreleased product, don't take that estimate as a gospel - besides you're comparing a full-die AMD card vs. an hilariously crippled last-generation product. The full TU117 wouldn't fare as poorly. This graphics card is a GT 1030 replacement, if the name didn't make that abundantly clear by now. These are intended to add multimedia support to computers without, or obsolete integrated graphics, at best very light gaming.

Navi 24 is certainly faster, but it's an incompetent HTPC GPU due to its poor display engine (inability to drive more than two displays) and limited support for media handling (no encoding capabilities whatsoever, limited decoding support). Pick your poison, do you want to play games or do you want multiple display-outs and the ability to transcode and watch movies? If it's the latter, the 1630 will be a better product to own.



It is but an assumption but I feel it's a safe one to make, that this would carry the same NVENC/NVDEC capabilities of the GTX 1650. It's the same die, and it's being marketed as a GTX, not a GT. Besides, being slower, NVIDIA needs this advantage against AMD, especially since Navi 24 inherently lacks the capability to do this even in the 6500 XT - the hardware simply cannot do it.

You have to compare on a dollar to dollar basis, not using some illogical excuses.

ARFYou have to compare on a dollar to dollar basis, not using some illogical excuses.

There's nothing "illogical" about what I said. People on this segment don't care about raw performance, they care about versatility, stability and low power consumption. They want something that can be installed, set up and will just work, often unattended for an extended period of time. This is the kind of graphics card you throw on a child's computer or on your movie rig in the living room.

As for pricing, AMD's low end offering is unsuitable for the HTPC market. Even the GTX 1050 provides a better deal there, so they have no product to compete at present. The pricing mentioned in the OP is not final as this graphics card has not yet been released, and even if it is that high, I bet many are willing to buy it because it will suit their needs. You're well aware that this applies to multiple segments including at the very high end for a similar reason, the 6900 XT games perfectly well, why would anyone buy a 3090? Because the 3090 offers you more features for your money, even if it doesn't offer you more performance.

ARFOverkill, like lipstick on a pig.

Not really. It's just a small aluminium block. Even with those two fans, it probably won't cool very well.

The cooler on my Asus TUF 6500 XT... now, that's overkill. :cool:

This is a response to Intel ARC coming out meant to fight their flagship card.

r9This is a response to Intel ARC coming out meant to fight their flagship card.

Hi,
Intel arc is a flagship ?

Seems more like a dinghy to me :cool:

Simply use the iGPU Radeon of an Ryzen:


AMD Ryzen 7 5700G Review: Fastest Integrated Graphics Ever | Tom's Hardware (tomshardware.com)

ThrashZoneHi,
Intel arc is a flagship ?

Seems more like a dinghy to me :cool:

It is ... as much as 1630 is. :D
Can you imagine how hard it will be to show how much slower Intel ARC is compared to nvidia next gen 4xxxx, the Intel ARC performance graph will be hair thin so they released 1630 ;)
:D

Dr. DroAs for pricing, AMD's low end offering is unsuitable for the HTPC market. Even the GTX 1050 provides a better deal there, so they have no product to compete at present.

I agree with the rest of your post, but this is a common misconception. The only thing the RX 6400 lacks for HTPC use is AV-1 video decode, which isn't there on the 1050, either. It's only there on bigger Navi 2x cards, Turing and Ampere. In fact, the 1050 is a lot slower in most cases (I've tested it with the Ti version) and consumes more power. I actually have a 6400 in one of my HTPCs, and it's doing great.

With that said, the only way the 1630 can compete with anything is if it comes in low profile, no power connector forms, and supports AV-1 decode.

ThrashZoneHi,
Intel arc is a flagship ?

Seems more like a dinghy to me :cool:

The only Intel ARC card released (and in limited quantities) in the Chinese market thus far is the A380 - which is their lowest-end design, with drivers in very early stages of development. I wouldn't write Intel off just yet, they may yet overcome the hurdles and present us with a good product. Simple driver updates have yielded extreme performance improvements in that GPU:


Improvements like that are unheard of in anything that's even remotely cooked, this card is raw ;)

ARFSimply use the iGPU Radeon of an Ryzen:

AMD Ryzen 7 5700G Review: Fastest Integrated Graphics Ever | Tom's Hardware (tomshardware.com)

Not everyone owns an AM4 platform and/or a motherboard with multiple display output ports. So between having e-waste on your hands plus spending hundreds of dollars on a motherboard, CPU and compatible memory or buying a dedicated graphics card, I guess the choice is clear as day. ;)

AusWolfI agree with the rest of your post, but this is a common misconception. The only thing the RX 6400 lacks for HTPC use is AV-1 video decode, which isn't there on the 1050, either. It's only there on bigger Navi 2x cards, Turing and Ampere. In fact, the 1050 is a lot slower in most cases (I've tested it with the Ti version) and consumes more power. I actually have a 6400 in one of my HTPCs, and it's doing great.

With that said, the only way the 1630 can compete with anything is if it comes in low profile, no power connector forms, and supports AV-1 decode.

I get what you mean, although, the 1050 would be able to do video transcoding in a box, the 6400 won't. Some people... kind of want that on a brand-new HTPC-oriented product. ;)

I doubt the 1630 will do AV1 support, I believe that was added to NVDEC with Ampere only and this is a TU117 part.

Chrispy_Has @W1zzard broken the review embargo and spoiled "the surprise" by uploading his review data to the database too early

Of course not, this is estimated by the same algorithm that we've been using for years, which usually works better than I expected when I wrote it

Dr. DroNot everyone owns an AM4 platform and/or a motherboard with multiple display output ports. So between having e-waste on your hands plus spending hundreds of dollars on a motherboard, CPU and compatible memory or buying a dedicated graphics card, I guess the choice is clear as day. ;)

Not everyone but most people. AM4 is the best PC platform ever released and the bestseller.

This card is oriented at those large stores for everything - like ready pre-assembled PCs which the average joes will buy "for the first time", and after that will be disappointed by the terrible performance. Because those people want to play modern games, which this clearly isn't able to offer.

ARFSimply use the iGPU Radeon of an Ryzen:


AMD Ryzen 7 5700G Review: Fastest Integrated Graphics Ever | Tom's Hardware (tomshardware.com)

Even if you own an AM4 system, upgrading to a 5600G is more expensive than the 1630 is probably going to be. AMD could have nailed it by releasing the 5300G on the DIY market, but they didn't for some reason.

RX 6400 LP exists? beats out the 1650 already and needs no PCIE, what am I missing?

DoLlyBirDRX 6400 LP exists? beats out the 1650 already and needs no PCIE, what am I missing?

Yes it does, in PowerColor and Sapphire flavours (look at my signature). ;) The only thing you're missing is crippled performance in some games with PCI-e 3.0 and older motherboards.

Vayra86A dual slot and you still need power from pcie for this utter piece of junk?!

What the hell is the market for this? Russia? There's even a reintroduction of DVI on this baby, man I had half expected a VGA next to it.

bet it comes with DVI/Dsub adapter and dual molex to 6 pin for power. if it came with a CD/game . . .

Yet another card based on dated tech. No thanks and i all ready have a gtx 1650 lp card.

If any thing, bring a rtx based card at least instead of this bs by bringing new cards on old tech. It's tiresome.

TomgangYet another card based on dated tech. No thanks and i all ready have a gtx 1650 lp card.

If any thing, bring a rtx based card at least instead of this bs by bringing new cards on old tech. It's tiresome.

How is Turing old tech? Is it the fact that it's 12 nm TSMC instead of 8 nm Samsung? Nearly everything else is identical in the two architectures.

AusWolfHow is Turing old tech? Is it the fact that it's 12 nm TSMC instead of 8 nm Samsung? Nearly everything else is identical in the two architectures.

Many years have passed since its inception? :confused:
12nm->7nm->5nm->3nm?

Apple produces 3nm chips right now. Where are we?
I understand that the mining craze and the coronavirus pandemic made many people accept unnatural solutions, but I think both are already over.

ARFMany years have passed since its inception? :confused:
12nm->7nm->5nm->3nm?

Apple produces 3nm chips right now. Where are we?
I understand that the mining craze and the coronavirus pandemic made many people accept unnatural solutions, but I think both are already over.

You can't compare an Nvidia GPU to Apple chips. Nvidia only works with 12 nm TSMC and 8 nm Samsung nodes. Apart from this, Turing offers the exact same feature set as Ampere, with similar efficiency. The fact that TU117 doesn't have Tensor and RT cores doesn't matter for the target audience. Nor does the fact that it's 12 nm and not 8. In fact, I don't care, either. My Intel CPU is 14 nm, so what? It's still good enough.

I measure computer hardware in FPS, seconds and Watts. Not in nm.

AusWolfYou can't compare an Nvidia GPU to Apple chips. Nvidia only works with 12 nm TSMC and 8 nm Samsung nodes. Apart from this, Turing offers the exact same feature set as Ampere, with similar efficiency. The fact that TU117 doesn't have Tensor and RT cores doesn't matter for the target audience. Nor does the fact that it's 12 nm and not 8. In fact, I don't care, either. My Intel CPU is 14 nm, so what? It's still good enough.

I measure computer hardware in FPS, seconds and Watts. Not in nm.

Nvidia is right now either in full production of 4 nm chips or only some weeks from commencing!

ARFNvidia is right now either in full production of 4 nm chips or only some weeks from commencing!

That's cool, but what value does that add to the current low-end GPU market?

AusWolfThat's cool, but what value does that add to the current low-end GPU market?

Do you want to convince everyone that it's stagnation and there will never be anything faster than GTX 1050 / GTX 1630 / GTX 1650 / RX 5500 XT / RX 6500 XT / RX 6400 in that low-end market?

If you don't care about performance, then simply buy integrated Radeon as in the Ryzen G-series APUs.

ARFDo you want to convince everyone that it's stagnation and there will never be anything faster than GTX 1050 / GTX 1630 / GTX 1650 / RX 5500 XT / RX 6500 XT / RX 6400 in that low-end market?

There will be, but currently there isn't. If I need a GPU, I need it now, not two years later.

ARFIf you don't care about performance, then simply buy integrated Radeon as in the Ryzen G-series APUs.

As in the £250 5600G together with a £100 motherboard and potentially with new RAM as well? No, thanks.

Nm is actually very important because it will make the chip smaller, cooler and you won't need any fans for it.
That Turing if shrunk to 4 nm would be a cheap and passively cooled low-profile discrete card.

ARFNm is actually very important because it will make the chip smaller, cooler and you won't need any fans for it.

Yeah, look at any Ampere card. They're so advanced that even the smallest 3050 has a power connector and fan(s).

ARFThat Turing if shrunk to 4 nm would be a cheap and passively cooled low-profile discrete card.

It still can be if its factory power target is configured right.