I still remember that before the Turing graphics card evaluation was lifted, TechPowerUp, the research and development company of the GPU-Z graphics card recognition tool, broke the news that two different hardware IDs appeared under the same product model, and the specific core code is that there is a difference of "A" and no "A". For example, TU-106-400 is a non-overclocked version core, and the frequency runs at the default frequency of NVIDIA; TU106-400A is an overclocked version core, and the frequency can be higher than the default frequency. To be honest, Turing graphics cards have been released for more than a month, and I have never seen any graphics cards that use the non-overclocked version of the core. I really want to know what the difference is, and whether it will be another "IQ detection card". This time we got the rare EVGA RTX 2070 Black Gaming graphics card, which uses the TU106-400 non-overclocking version core, so what about the performance gap?
RTX 2070 Founders Edition is not available for sale on the market. You can see that the first batch of RTX 2070 graphics cards are all non-public versions with slightly "magic modifications", either the PCB modifications or the TU106-400A overclocking version core is adopted. It was also a coincidence that the discovery that the EVGA RTX 2070 Black Gaming graphics card uses TU106-400 non- overclocking version core is the result of a coincidence, because the EVGA graphics card currently only has a formal sales channel in JD.com in the Chinese market, while the RTX 20 series graphics card has not yet been launched. We have created a basic version of the EVGA RTX 2070 Black Gaming graphics card through special channels.
EVGA RTX 2070 Black Gaming graphics card is indeed the most common version, belonging to the Gaming series, pure public version PCB solution, the materials are also consistent with the public version formula, and it does not have lights, which is extremely simple and unpretentious. There are also mid-range XC series and high-end FTW series. The horizontal design of most manufacturers has also emerged on the packaging of
. The vertical packaging design is indeed quite outstanding. The upper right corner is deliberately made into a cut-out edge, which has increased the difficulty of packaging.
Through the preview, we learned that the appearance of the EVGA RTX 20 series graphics cards has been greatly improved, and all of them use black acrylic air shields. This design reminds me of the graphics cards of Yingchi many years ago. Is EVGA making the "Exploration Edition" Turing graphics card? However, acrylic is not scratch-resistant, so everyone should take care of it. If there are scratches, it will not look good.
EVGA in the past insisted on creating dual-slot graphics cards, but due to the increasing core, it is difficult to eliminate the heat as soon as possible. In addition, I have tried "over-popular", EVGA is creating some 2.75-slot graphics cards, but the EVGA RTX 2070 Black Gaming graphics cards we got still maintain the original dual-slot.
translucent air guide hood does have a unique flavor, and it is brown/black acrylic, which will bring a better visual impression, but it also makes the graphics card feel too cheap and can still maintain that "b-shaped", and you can even clearly see the internal structure of the radiator fins.
EVGA is good at creating details. You can see that the two nameplates "EVGA" and "GEFORCE RTX 2070" are just right, indicating the identity of its own RTX Turing graphics card.
The top of the graphics card is printed as always with the words "EVGA GEFORCE RTX 2070". If it is the RGB version, it will be luminous.
What is more depressing is that although the Gaming series graphics cards are entry-level, they are not equipped with a full-coverage back panel, right? This is already 2018, and graphics cards without a back panel are really rare.
EVGA is a small back panel to block it in order to cover the extended heat dissipation fins, and there is also a hollowing of EVGA on it.
fans are also exclusively customized by EVGA. Don’t ask me why I know, because the fan blades are densely packed with EVGA LOGO, and I’m going to be experiencing my intensive phobia. EVGA says the fan blade has a special “E” letter design that reduces wind-cut noise by 4%.
RTX 2070 Founders Edition adds the previously cancelled DVI-D interface. The same is true for the EVGA RTX 2070 Black Gaming graphics card, which collects all digital video output interfaces, including DisplayPort×2, HDMI×1, USB-C×1, DVI-D×1.
Disassembly:
Although it is also a public version of PCB solution, it is much better to disassemble than the RTX 2070 Founders Edition that kills people. The standard Phillips screws can be disassembled after being screwed off.
EVGA heat dissipation solution is also simple and crude, with large-area heat dissipation fins, pure copper nickel-plated base, three large-diameter heat pipes flattened with them, and the double-slot design, is it necessary to give a question mark to its temperature performance?
GDDR6Video memory , the VRM power supply part is responsible for heat dissipation, large-area thermally conductive silicon pads, and can also enhance PCB strength.
is said to be a PCB, and the 8Pin power supply interface does not have flying wires, which looks much more harmonious and tidy.
EVGA RTX 2070 Black Gaming graphics card core is confirmed to be TU106-400, it really does not have A, and the performance is worth looking forward to. The 8GB GDDR6 video memory surrounds the upper side in a 2-4-2 way, from Micron D9WCW series, 14Gbps.
EVGA RTX 2070 Black Gaming also uses 6+2 phase power supply, and the core power supply is 6 phases. MosFET has also been replaced with ON Semi's NCP 302155, with a maximum throughput flow of 55A.
video memory MosFET is also different, it is ON Semi's dual N-channel MosFET FDPC 5018SG.
The last public version of the PWM control chip was not clear, but this time I saw it clearly, and it is still an independent and shutdown eight-channel PWM controller uP9512.
test platform:
Basically all RTX 2070 graphics card reviews are based on this test platform. Intel Core i7-8700K processor is paired with Gigabyte Z370 AORUS Ultra Gaming 2.0, dual-channel Gigabyte AORUS DDR4 3200MHz 8GB×2 memory, Intel 600p 512GB M.2 SSD, and the gaming disk is an external Seagate Cool Fish 2TB mechanical hard disk. For some other electromechanical products, please refer to the list.
First check the EVGA RTX 2070 Black Gaming specifications and parameters through GPU-Z. Wow, the standard is the NVIDIA RTX 2070 "public version frequency", the same as the most standard frequency given by NVIDIA. The default frequency is 1410MHz, the boost frequency is 1620MHz, and the equivalent video memory frequency is 14Gbps. Everything is in the most standard state.
Default TDP is 175W, the upper limit can be adjusted to 114%, that is, 200W, and the maximum temperature can be adjusted to 88℃. Except for the low power consumption of TDP, there is actually not much difference.
benchmark performance test:
According to convention, we all use 3DMark as the graphics card benchmark test, namely 3DMark FireStrike, FireStrike Extreme, FireStrike Ultra, TimeSpy, and TimeSpy Extreme. The specific scores have been shown in the chart below. If you want to know more about the performance of the game frame rate, you can click here to transfer it to the first review, which will give you more complete test data for reference.
's most exciting time has come. We will finally reveal the performance gap between overclocking core and non-overclocking core. Here we will compare the RTX 2070 Founders Edition. TU-106-400 VS TU-106-400A showdown begins!
As expected, the TDP is not high enough and the boost frequency is a disadvantage. The EVGA RTX 2070 Black Gaming 3DMark scores are all lower than the RTX 2070 Founders Edition, so the core performance of the non-overclocked version will indeed be slightly affected. If the price is the same, I don’t need to talk about which one to choose, right?
NVIDIA is actually a bit unauthentic in distinguishing cores. It is difficult for consumers to identify which of the overclocked version is the core of the tens of millions of models. It is necessary to do their homework well, but even if you buy it, you don’t have to be angry, because the performance gap is not large, and the market share of these core graphics card products is very low.
EVGA Precison X1 software display:
I believe everyone has heard of some unknown and unclear relationship between the EVGA brand and NVIDIA. It is true that after the release of the RTX 20 series graphics card, NVIDIA's primary recommended overclocking software and displaying the NVIDIA OC SCANNER API functions are all completed by EVGA Precison X1 software. Isn't this the treatment of a "own son"?
EVGA Precison X1 software is open to all brands. There is no exclusive EVGA graphics card, and it can be used after installation.
Under the newly designed UI, EVGA Precison X1 is more intuitive and clear, without fancy functions. The above is the overclocking interface and fan control. If you switch to the second interface, you can also draw the fan operation curve by yourself.
There is also a graphics card status monitoring interface, and it can even be displayed on the game OSD.
and VF CURVE Tuner, which is NVIDIA's OC Scanner function.
If your EVGA graphics card has RGB lights, it can be controlled through the LED interface, including the lights of the NVLink bridge, which is very awesome!
Overclocking test:
Since the performance of the non-overclocking version is not as good as the core of the overclocking version, the overclocking ability is definitely weak. After all, the core has been selected by NVIDIA in advance and is "branded", so overclocking is even more difficult. At least that's what I thought before the test.
NVIDIA's OC SCANNER is estimated to be able to lower the frequency upper limit depending on the recognition of the core code. If the frequency upper limit is adjusted, the RTX Turing graphics card will have less bonus overclocking. In this way, can the TU-106-400 core still enjoy overclocking?
is shocking, can overclock, and play like that! After unremitting efforts, the EVGA RTX 2070 Black Gaming graphics card was overclocked to 1570MHz, and the graphics memory was overclocked to 16Gbps, which is equivalent to core +160MHz and video memory +1000MHz.
etc. The non-A suffix that was said was too unchanged, is this a scam? Or did you meet a physical divine eagle? Anyway, the performance improvement is between 10-12%, far exceeding the TU106-400A core non-public graphics card we tested before. The improvement is large, partly because the default score of the EVGA RTX 2070 Black Gaming graphics card is relatively low, but after overclocking, the results surpass some non-public RTX 2070. It is really a magical thing.
Who says that the TU106-400 core cannot be overclocked? Please drag it out and hit 100 large boards.
Oh, right, after the TU106-400 core is overclocked, the boost frequency is relatively low, and it is difficult to go to 2GHz. But if it breaks through, it will be suppressed soon. The TU106-400A core can be maintained for a long time, but the results seem to be similar between the two.
Temperature test:
indoor temperature is 24-25℃, the box is sealed throughout the temperature test, and the test chassis is the coolest MasterCase H500 chassis. The standby temperature is recorded for 10 minutes after entering the system, and the full load temperature is recorded for 3DMark Fire Strike stress test for 10 minutes. In order to allow everyone to see the temperature changes of the graphics card intuition, the GPU-Z Log to file function was specially used to output all graphics card data, and a line chart was collected in 10 minutes to show it to everyone.
EVGA RTX 2070 Black Gaming graphics card is also quite awesome. The fan stops in standby state, which can ensure absolute quietness. The temperature is finally stable at 39℃, which is slightly hotter than other homes 7-10℃, but it can have zero noise, which is a good choice. Under pressure test, the dual-slot, three-heat pipe, and dual fans can even lower the temperature and press it to 61℃. This is amazing. Some three-fan graphics cards may not be so cool.
power consumption test:
Through our dedicated graphics card power consumption testing instrument, we can accurately measure the watts of the graphics card PCI-E and external power interface respectively. The maximum power consumption of the graphics card is obtained in the 3DMark Fire Strike stress test, and the standby power consumption is recorded for 1 minute after entering the system.
As an over-standard "public version parameter" graphics card, the average standby power consumption of the EVGA RTX 2070 Black Gaming graphics card is only 11.8W, and the average full-load power consumption is 177.3W, which is also quite standard.
summary:
found that the EVGA RTX 2070 Black Gaming graphics card uses the TU106-400 non-overclocking version core. It is an accident that the difference between it and the TU106-400A overclocking version core is just enough to check the difference between it and the TU106-400A overclocking version core. However, it was unexpectedly discovered that although the performance was not as good as the latter by default, after some overclocking operation, the situation reversed. The overclocking performance of the non-overclocking version core of the is stronger than the overclocking version core. Although it cannot be ruled out that it is an individual problem, it is enough to let us put down our tinted glasses to see the standard frequency graphics card product. It may also be a piece of rough jade, but it has not been polished and can emit dazzling light after your manual polishing. Although the
EVGA RTX 2070 Black Gaming graphics card is an entry-level series product, EVGA is not careless, simple, unique and ingenious appearance design, good heat dissipation ability, and the same materials as the public version. If the price can reach the minimum recommended retail price of 3,999 yuan recommended by NVIDIA, this is definitely a worthwhile RTX 2070 graphics card.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-