Friday , May 24 2019
Home / zimbabwe / Budget Turing ($ 149 today)

Budget Turing ($ 149 today)



NVIDIA's GeForce Turing product stack finally came out in eight months after it all started. This morning the company presents the GeForce GTX 1650, the latest inexpensive GeForce Turing video card. The latest member of the GeForce family of NVIDIA's latest architecture, the latest member of the GeForce family, is available for $ 149 at $ 149. Low-cost 1080p video game cards are available at a fraction of the cost.

In a very traditional Nvidia fashion, the Turing launch was one-on-one. Within two months of launching four RTX 20 Series cards, NVIDIA focused on filling the back of the product stack. At the heart of this is NVIDIA's GPU design variant, the TU11x series. Turing Minor is designed to be easy to produce chips while maintaining a smaller, more important Core Turing architecture. Provides dedicated ray tracing (RT) core and artificial intelligence centered tensor core. The end result of this bifurcation was the GeForce GTX 16 series. The GeForce GTX 16 series is designed with a more sophisticated and boring set of GeForce cards.

To date, the GTX 16 series has been made up of GTX 1660 (vanilla) and GTX 1660 Ti only GTX 1660 family. Both were based on the TU116 GPU. Today, however, the GTX 16 series is expanding with the introduction of the GTX 1650 and the new Turing GPU, NVIDIA's Junior size card, the TU117.

When the GeForce GTX 1660 Ti and the basic TU116 GPU first saw NVIDIA's mainstream product plan, the GeForce GTX 1650 became much more pedestrian. The underlying TU117 GPU is a smaller version of the TU116 GPU for all practical purposes, reducing all resources while maintaining the same set of Core Turing features. NVIDIA from the TU116 shaved one third of the CUDA core, one third of the memory channel, one third of the ROP, and left a smaller, more manufacturable GPU for this low-margin market. With a size of 200mm2 and a 4.7B transistor, the TU117 is not just a chip. In fact, it's exactly the same die size as GPU6, the GPU in the heart of the GeForce GTX 1060 series, so you can get an idea of ​​how slowly performance and transistors are slowly dependent on low-cost products over the past few years.

Anyway, the TU117 will be applied to many NVIDIA products over time. But now we start with the GeForce GTX 1650.

NVIDIA GeForce specification comparison
GTX 1650GTX 1660GTX 1050 TiGTX 1050
Cuda Core8961408768640
ROP32483232
Core clock1485 MHz1530MHz1290 MHz1354 MHz
Boost clock1665 MHz1785 MHz1392 MHz1455 MHz
Memory clock8Gbps GDDR58Gbps GDDR57Gbps GDDR57Gbps GDDR5
Memory bus width128 bit192 bits128 bit128 bit
VRAM4GB6GB4GB2GB
Single precision.3 TFLOPS5 TFLOPS2.1 TFLOPS1.9 TFLOPS
TDP75W120W75W75W
GPUTU117
(200 mm & lt; 2 &
TU116
(284 mm2)
GP107
(132 mm & lt; 2 &
GP107
(132 mm & lt; 2 &
Number of transistors4.7B6.6B3.3B3.3B
buildingTuringTuringpascalpascal
Manufacture processTSMC 12nm & quot; FFN & quot;TSMC 12nm & quot; FFN & quot;Samsung 14nmSamsung 14nm
Release DateApril 23, 2014March 14, 201410/25/201610/25/2016
Release Price$ 149$ 219$ 139$ 109

Just the bat, the GTX 1650 no I use a fully activated TU117 GPU. Compared to the entire chip, the GTX 1650 incorporates the TPC and the chip has lost two SM / 64 CUDA cores. The end result is that the GTX 1650 is a very rare case where NVIDIA does not take the very first step in bat. The company is essentially sandwiched.

It is somewhat difficult to place the GTX 1650 within NVIDIA's historic product stack. Officially became a successor to GTX 1050. The GTX 1050 was a similar cut-down card by itself. However, while the GTX 1050 was priced at $ 109, the GTX 1650 was priced at $ 149 and the power generation price was up 37%. As a result, if you think that the GTX 1650 is much more like the successor of the GTX 1050 Ti, you can be forgiven if you think the $ 149 price tag is very similar to the GTX 1050 Ti's $ 139 release price. Either way, generation-over-generation, Turing cards are more expensive than the Pascal cards they replace, and the low price of these budget cards actually boosts the gap.

Then the GTX 1650 is loaded with 896 CUDA cores and spread to two GPCs. In fact, it's not much bigger than the GeForce GTX 1050 series, but Turing's architectural changes and effective increase in graphics efficiency mean that a small card should pack more punches than originally shown on paper. The CUDA core itself offers slightly lower clock speeds than usual for Turing cards, but the reference clock GTX 1650 is only improved to 1665 MHz.

There are 32 ROPs to complete the package, which is part of the card's four ROP / L2 / memory clusters. This means NVIDIA is supplying cards to a 128-bit memory bus that is paired with GDDR5 memory clocked at 8 Gbps. Conveniently, this provides 128 GB / sec of memory bandwidth, which is about 14% higher than the next generation GTX 1050 series card. Thankfully, NVIDIA has not made much effort to increase the memory capacity of other Turing cards, but the same is true for the GTX 1650. The minimum size is 4 GB, not 2 GB found on the GTX 1050. It's not like that. 4 GB was particularly widespread in 2019, but the cards were as desperate as their predecessors.

Overall, the GTX 1650 on paper was set to deliver about 60% of the following card performance on NVIDIA's product stack, the GTX 1660. In fact, these two are a little closer. GPU performance tuning is done with the & # 39; We are now watching the baseball stadium until we can actually test the card.

On the other hand, the smallest member of the GeForce Turing stack is the lowest power in terms of power consumption. NVIDIA currently has 75W (or less) GTX xx50 cards for generations, and the GTX 1650 continues this trend. This means that cards that operate at least at the NVIDIA reference clock do not require an additional PCIe power connector and can only receive card power from the PCIe bus. This meets the need for a card that can be plugged into a low-power system where a PCIe power cable is not available or a power-hungry card is not suitable. This also means that individual video cards are not as popular as they were for HTPC, but for those looking for HTPC builders, the GTX 1650 will be a replacement for the GTX 1050 series in that market.

Reviews, product positioning and competition

Let's talk about product location and hardware availability as we move gears to business issues.

The GeForce GTX 1650 is NVIDIA's hard drive. In other words, cards will be shipped today from retailers and OEM systems. For low-end NVIDIA cards, there are no reference cards or reference designs to mention, so NVIDIA's board partners perform their own work through their product lines. Specifically, the factory that provides more performance includes overclocked cards, but an external PCIe power connector is required to meet the card's greater energy needs.

But this is a very authentic (though not entirely covert action) NVIDIA has decided not to allow the media to pre-test the GTX 1650 card.. Specifically, NVIDIA has withheld the drivers needed to test the card. In other words, you could have secured the card in advance, but you would not have it. We have cards on hand and will write reviews in a timely manner. But you do not have any real experience with the GTX 1650 card than you do.

NVIDIA has always handled low-end card releases at a lower price than high-end products, and so is the GTX 1650. In fact, the release of this generation is not particularly important. Since NVIDIA chose to notify you via e-mail, there are no pictures or press decks. At this point there is little need for a wide fanfare. The Turing card and the Turing architecture / feature set are over-worked at this point. New GPU-based cards rarely start without reviewers having an initial crack. From it. This is a good reason. Reviewers analyze cards and their performance through neutral third-party analytics. Therefore, in general, cutting a reviewer is not the buyer's biggest concern. And when will this be a red flag?

Anyway, at this point Turing is a known quantity, while the buyer is suggesting to wait a week and write a review. As mentioned earlier, the paper specification places the GTX 1650 at about 60% of the performance of the GTX 1660, and the actual performance will be slightly higher. NVIDIA basically adopts cards as an upgrade to GeForce GTX 950 and its generation of AMD-compliant products, and there was the same upgrade cadence difference seen in the rest of the GeForce Turing product line. NVIDIA says the performance is twice as fast (or more) than the GTX 950, which should be easy to achieve.

While we wait for the cards to run the benchmark, the GTX xx50 series cards mean 1080p and compromise cards, and the GTX 1650 almost anticipates what it is based on. We saw the GTX 1660. The GTX 1650 should be able to run 1080p games with maximum image quality considering DOTA2 and so on. However, in more demanding games, there is a 1080p playable frame rate. However, one of the benefits here is that with 4GB of VRAM you should have little difficulty with newer games than the 2GB GTX 950 and GTX 1050 do.

Finally, as for competition, AMD is on the tail end of the Polaris-based Radeon RX 500 series, so this will be against the GTX 1650. AMD is working hard to install the Radeon RX 570 8GB compared to the GTX 1650. It's a very interesting battle. Based on what we saw on the GTX 1660, the RX 570 will work quite well with the GTX 1650, and the 8GB VRAM will be decorated on the cake. However, we can not be sure that AMD and its partners will be able to keep the price of an 8GB card below $ 149, in which case the competitor can be a 4GB RX 570 instead.

Ultimately, AMD's position does not match the GTX 1650 to function or power efficiency, but keep in mind that the RX 570 consumes almost twice as much power. Price and price can be adjusted to performance. When AMD wants to keep this line right now, it will be an advantage for AMD in terms of price / performance of current generation games. To see how advantageous it is, you will need to benchmark the GTX 1650, so keep watching.


Source link