Power limitation of 3060 graphics card
Yes
several new procts of NVIDIA graphics card have been greatly improved at the press conference. The performance of rtx3070 graphics card has surpassed rtx2080ti, and rtx3060 over 2070 is easy to achieve. The performance of rtx3060 will be 30% higher than rtx2060, similar to rtx2070
rtx3060 is equipped with 12gb video memory, which is higher than rtx2080ti with 11gb video memory and rtx3080 with 10GB video memory. In terms of model, rtx3060 uses rtx2060 graphics card, which also adopts 106 architecture and 192bit bit width
because the graphics card uses a different core architecture from rtx3060ti (rtx3060ti card uses ga104 Architecture), there is still a certain gap in performance
extended data:
3060 graphics card is not very good in the actual performance at present. The performance of graphics card is between rtx2070 and rtx2060super, which is slightly inferior to 2070 in performance, about 10-15% higher than 2060super. I have to admire Lao Huang's skill. You don't even think of it as a 3060, but as a 3050ti
generally speaking, the improvement of the 3060 is limited compared with that of the previous generation, so Xiaobian doesn't recommend you to consider it very much. After all, the price of the current graphics card is rising very high, so we don't recommend you to buy the 3060 at this stage. On the one hand, it's not worth the price. On the other hand, it's the reverse upgrade of the performance. Besides, the price of more than 2500 yuan is still a little too expensive, Besides, even if it's the original price, you may not be able to buy it, so at this node, if you buy it, you will suffer a loss. At least the overflow price is about 1000
the performance sling has to be increased without price increase, which makes many rtx2080ti players cry and faint in the toilet (including me). As the saying goes, the back waves of the Yangtze River push the front waves, and the front waves hang on the beach. On the small yellow croaker, the rtx20 series resale scene is full of laughter: 2080 buy 2080ti, small theater, 50 years into G Jun, and so on. But after all, the rtx20 series is really not worth cutting hands? No, after all, like the new iPhone, a lot of people "pick up" the previous generation's procts after the new generation's release. No, recently they helped people install the machine, and they really added a rtx2060super<
Proct unpacking:
the configurations are all in the diagram. I'm too lazy to list the configuration table, which is about r53600xt + b550 + 32g memory + rtx2060super
this rtx2060super is Elsa magic thunder. It may be strange to see this brand of DIY cute new players, but in the early years, it was a first-line brand, and now the first-line Jijia was only OEM for it. However, this brand has been acquired and re opened now, but the market is no longer the pattern of that year. In the past, big men can only take the cost-effective route. This video card is a brand new proct from JS, because the original channel price was not high, and JS was afraid to smash it after seeing the 30 release, so it was transferred to me at a good price. There is no shortage of awesome prices, right? I believe that if there is a 2080 selling 2080ti now, it will be a second
the CPU uses r53600xt, and the motherboard with it is naturally the best b550. Some people say why not choose the cheaper b450. Well, in order to perfectly upgrade the fourth generation of sharp dragon, and the b450 is not as good as pcie4.0. Although the solid-state of pcie4.0 is still expensive at present, when the b550 is cheap, it will not be able to fully open the performance? It has to be said that the life cycle of b550 is definitely much longer than that of b450
the motherboard of MSI MAG b550 Tomahawk missile is designed with ATX board type. The materials used are fairly standard. After all, the price is here
10 + 2 + 1 power supply is a bit of a killer for r53600xt, but better materials are also excellent to deal with the fourth generation of sharp dragon. Maybe the new generation of sharp dragon will continue to play the core war? This is the main board power supply demand is higher
with four memory slots, the maximum capacity is 128GB. When a-xmp mode is turned on, the maximum frequency is 4866mhz +. However, as a fan of Lin University, I firmly believe that this board's memory overclocking of 5000mhz is no problem
the charm of b550 lies in the fact that the first m.2 interface near the CPU and the pcie4.0 slot support pcie4.0. In addition to using the cheaper pcie4.0 solid state in the future, rtx3060 can also support pcie4.0 perfectly. Do you know why I choose amd platform this time
the I / O baffle of the graphics card is integrated on the motherboard, which saves a lot of trouble. Before the installation, I used to find the I / O baffle from the box. This B550 also supports al network cards, which is really awesome, with a 2000 trillion Gigabit configuration.
the second channel m.2 interface is also equipped with heat sink, which supports PCI 3.0 x16, which is better than some b550 second channel m.2 interfaces that are not full speed 3.0 on the market
the memory uses Kingston fury DDR4 3733 16GB * 2, and the frequency is 3733mhz. It belongs to the thunderbolt series of hackers. Without RGB, it is more cost-effective, with PNP characteristics, and can achieve the ideal frequency without setting. Of course, you can also enter the BIOS for XMP one click overclocking. The outer layer of metal heat dissipation vest can heat the memory particles. Another reason for choosing Kingston is its excellent compatibility and stability
the chassis uses the josper u5s, which is equivalent to the vertical mounting version of the U5 graphics card. Metal wire drawing shell, front two USB3.0 interfaces, audio interface, one big and one small is the power key and restart key. Now many chassis abandon the restart key, which is really honey operation
both sides of the chassis are made of glass. It is recommended that it is safe to remove and toss when installing the chassis<
stproof net
at first, I thought this case was a stuffy can. I used to play with the vertical air ct, which has two 12cm fan installation positions, can also be used for 240 water cooling, and the left side is the power supply compartment
three 12cm fans can be installed at the bottom of the chassis, and 360 water-cooled fans can also be installed. On the left side is the vertical mounting bracket of the graphics card
the ADT adapter attached to the box can support pcie3.0 x16 speed at most. It's rash to see this. You may need to change the chassis to replace the new graphics card with pcie4.0 feature in the future
at the back of the chassis, there are three slots for the power interface and 12cm fan
in terms of power supply, it uses anatik ne650w, full mole design, 569 seconds to get it, 7 years to replace it, 80plus gold medal certification, plus all Japanese series capacitors. It's OK to be less than 1 yuan and 1W. Moreover, anatik is also the first line of electromechanical brand, and the quality makes me feel at ease. Helping others install their machines is my favorite choice. It saves me a lot of time and effort to replace them with new brands
did you find that the radiator has become a down pressure type? Because I saw the valet who used this case remind that water cooling is the best on this proct. Because the graphics card is installed vertically, if the air-cooled radiator is too wide, it may conflict with the graphics card. In order to avoid the trouble of second rework, he directly replaced it with down pressure cooling. The final installation effect is OK
performance:
r53600xt performance is OK, although compared with r53600, Rui frequency is only 0.1ghz higher, JD is 200 more expensive, but I quietly tell you that offline prices are similar, only less than 100 more expensive
3dmark performance overview
rtx2060s and 2K resolution high special effects are OK. There is no pressure, let alone 1080p resolution setting
summary:
it's not worth buying rtx20 series now. It depends on the price. After all, although rtx30 series is strong, rtx20 series is not so unbearable. Its combat effectiveness is still there. At least, in addition to its 30 big brothers, there is no rival in the market. Here, I want to feel sorry for AMD. With the collapse of rtx20 Series in the future, I believe that the a-card currently on sale will be hit hard. However, it's said that AMD's big Navi is coming. It's said that its performance can match the rtx3080. Let's wait and see
for the last generation of "series 2" dddr3, the performance of N card is divided from low to high, low-end gt210 to gt240, low-end gt250 and high-end gtx260. Note that 260 becomes GTX, not GT, but gtx270 and high-end flagship gtx280
for the last generation of "4 series" dddr5, the performance of n-card is divided from low to high, the low-end gt410 to gt440, the middle end gt450 and the middle end gtx460. Note that 460 becomes GTX instead of GT, the high-end gtx470 and the high-end flagship GTX480
for the last generation of "5 Series" dddr5, the performance of N card is divided from low to high, the low-end gt510 to gt540, the middle end GT550 and the middle end gtx560. Note that here 560 becomes GTX, not GT, the high-end gtx570 and the high-end flagship gtx580
for the last generation of "6 Series" dddr5, the performance of n-card is divided from low to high, the low-end gt610 to gt640, the middle end gt650, and the middle end gtx660. Note that 660 becomes GTX, not GT, the high-end gtx670, and the high-end flagship gtx680
for the last generation of "7 Series" dddr5, the performance of N card is divided from low to high, the low-end gt710 to gt740, the middle end gt750, and the middle end gtx760. Note that the 760 becomes GTX instead of GT, the high-end gtx770 and the high-end flagship gtx780
for the last generation of "9 series" dddr5, the performance of n-card is divided from low to high, the low-end gt910 to gt940, the middle end gt950, and the middle end gtx960. Note that 960 becomes GTX, not GT, and the high-end flagship gtx980
for the new generation of "10 series" dddr5, the performance of n-card is divided from low to high: the low-end gt1030, the middle end gt1050, and the middle end gtx1060. Note that the 1060 becomes GTX, not gt; the high-end gtx1070 and the high-end flagship gtx1080
for the new generation of "16 series" dddr5, the performance of N card is divided from low to high. Note that all of them are GTX, the middle end is lower than gtx1650, and the middle end is higher than gtx1660
the new generation of "generation 2" of "16 series" uses dddr6 (ddr6 is starting to appear here). The performance of N card is divided from low to high. Note that this is GTX, the middle end is lower than gtx1650, and the middle end is higher than gtx1660. The "16" series of ddr6 and ddr5 can't be distinguished from each other in terms of model. You can only check whether it is D6 or D5 from the parameters of the graphics card
the new generation of "20 series" dddr6, n card performance from low to high division, note that here are RTX, not GTX, and support light chasing effect. All models have become RTX, the middle end is higher than rtx2060, the upper end is lower than rtx2070, and the high-end flagship is rtx2080
summary:
1. See if there is an "X" in the model. Those without an X belong to the low end, while those with an "X" belong to the middle end or high end
2. Look at the numbers in the models. There are 1 to 4 in the model numbers, which are the low end, 5 in the middle end, 6 in the middle end, 7 in the high end, and 8 in the high end flagship. If there are 9 in the flagship, they are the top graphics cards in some series
3. The tail of the model is "Se", which is a simplified version of NVIDIA, that is, a weakened version. The suffix is "OC", which is the version modified by the brand manufacturer. It doesn't have to be weakened or enhanced. It depends on the parameters. The "Ti" is the enhanced version of NVIDIA, and the "super" is the vest changing version of NVIDIA. If the prefix of the model is "GTS", it is the enhanced version of "GT". So many other prefixes or suffixes are basically the models added by brother brand manufacturers after their own changes, and they are not NVIDIA's version< Conclusion:
(1) then RTX 3060 should be the middle end graphics card in "30" series, and its performance should be 20% to 25% higher than that of the previous generation "20" series rtx2060, which should be equivalent to that of rtx2070
(2) if it is RTX 3060ti, because ti is an enhanced version, its performance should be 50% to 55% higher than rtx2060 and 40% to 45% higher than rtx2060s. 30% to 35% higher than rtx2070 and 20% to 25% higher than rtx2070s. About 10% to 20% higher than rtx2080, equivalent to rtx2080s, and 10% to 20% lower than rtx2080ti
(3) the above are speculative performance, for reference only!
one is the interface supported by the motherboard. If it comes out, it is estimated that it supports the pcie4.0 standard. You are using 3.0 now. You can definitely use it, but the performance will shrink a lot. After all, the bandwidth gap is very big! One is power consumption. How much power is the minimum requirement It has nothing to do with CPU)
it is said that the power consumption is very low, and the performance of 3060 can reach 2080ti level, but it is all said by the media! I can't believe it. Just have a look! I don't know when it will be released
if you want to use this kind of graphics card, the motherboard is the key!!
with the performance of i7 8700 processor, it can meet the requirements of playing games at present, so there is no limit to using a single graphics card, and you can use rtx3060 graphics card.
