Chapter 608 NVIDIA and AMD
(The data is a bit too much. If you don’t like data, you can skip saving.)
“And the most important thing is that Jiuzhou Technology has committed a big mistake. The mistake is that they should not come out with the Titan processor, Loki X1 and Spirit Lander at once. If they come out step by step, in three years and one stage at a time, I They are definitely incomparable, but it is different now
Jiuzhou Technology has come up with too many high-tech products in a short period of time, which has caused panic in the market and forced us to join forces to deal with it. This is a very unwise move. Measures.
I dare to assert that if Jiuzhou Technology cannot produce better products in the future, it will become the second Nokia, and we will learn from it and then surpass it.
We will eventually take over the Daxia market again, and there are many architectural design patents for the processor. These are barriers that they cannot get around! ”
Lao Huang was eloquent and eloquent, and even made a prophecy that Kyushu’s technology was very strong, but it would also die because of its strength.
To be honest, his plan It’s a lie. Directors who don’t understand technology believe it because they do have Example, death due to strength.
After hearing the voice-over about learning technology, those executives and directors secretly nodded.
How can technology research and development be done? What’s called stealing? That’s called learning!
When Lao Huang was painting cakes and talking, he kept paying attention to the expressions of other people. Seeing that everyone's expressions were much more relaxed, he was relieved of his worries.
The words that came one after another were difficult to understand, such as "information technology and the metaverse" and "TSMC's lead in process orders", which caused everyone to burst into laughter: the atmosphere inside and outside the conference room was filled with joy.
Finally someone dared to ask in a low voice: "Huang Renxun, what you said is predictable, but do we have to unite with competitors like Pingguo and Intel?"
"Of course. Impossible!" Huang Renxun said hurriedly: "Discussing cooperation with Jiuzhou Technology. We will not relax for a day. As soon as Jiuzhou Technology relaxes, we can immediately prepare for joint venture business. At this stage, Jiuzhou Technology’s technology is still something we need to learn.”
See CEO Lao Huang. As rigid as Apple's Cook is, the directors' demeanor is more relaxed.
In this market environment, Western technology companies have also selectively used weapons that they think are OK.
NVIDIA RTX40 series graphics cards: manufactured using TSMC’s 5nm process, currently including five GPU cores, namely AD102, AD102, AD104, AD106 and AD107.
The AD102 with the highest specifications has up to 18432 CUDA cores, the AD103 has 10752 cores, and the remaining three models have 7680, 4608, and 3072 cores respectively.
For comparison, the current NVIDIA card king RTX3090 has 10496 CUDA cores. If the RTX4090Ti uses the full-blooded version of AD102, the number of cores will increase by 7936, and the performance will increase by at least half.
According to gossip, the gaming performance of RTX4090Ti is equivalent to that of Titan Processor C!
There is also "supporting news", that is, Intel will release a new 13th generation Core Raptor Lake processor and 700 series chipset motherboard at the end of the year. By then, it will assemble a 13th generation Core processor, 700 series motherboard, DDR5 memory, PCIe5.0M2 solid-state drive, coupled with Nvidia RTX40 series graphics cards, its performance can compete with Titan processors.
Intel and Nvidia are competing against Kyushu Technology. When AMD sees the opportunity coming, how can it be willing to give up the market?
AMD CEO Su Ma stood up and said: "Please shout the slogan of players! AMD YES!" Before the launch of the Titan processor, AMD launched the RX6000 series of graphics cards based on the RDNA2 architecture, including Radeon RX6800, 6800XT and 6900XT. Targeting NVIDIA's RTX3090, RTX3080 and RTX3070 respectively.
Although the graphics card is very good, AMD's game support, software support and support for Windows systems are still much worse than Nvidia's. Therefore, although the graphics card is very strong, it is slightly inferior in practical applications. Weak and insufficient production capacity!
However, under the strong pressure of the Titan processor, as well as Intel's big brother graphics card, NVIDIA 40 series, and the special environment of the mining disaster, AMD also had to come up with the "long-cherished" 7000 series graphics card using the RDNA3RX architecture
< br>Information It shows that the energy efficiency of the 7000 series graphics cards based on the RDNA3RX architecture will be increased by 50% again on RDNA2, and due to capacity expansion and technology upgrades, this generation of AMD's 7000 series graphics cards will still have sufficient cost-effectiveness and game optimization performance.
Of course, in the parameter table of data benchmarking, Jiuzhou Technology’s Titan processor also appears again.
Su Ma held the new graphics card with her right hand wearing a pigeon egg ring, and said confidently: "The reason why the running score performance is so high is because the 7000 series graphics card in my hand uses a new architecture. In addition, we also have a new technology. The AMD Radeon Pro SSG solid-state graphics card technology that previously had a built-in solid-state drive has begun to be gradually distributed to ordinary user products. "
AMD Su Ma was shocked by this gas stove. A group of people.
Putting the solid-state drive into the graphics card, this technology is too similar to the Titan processor!
Looking at the running scores and actual measurement data, the 7000 series graphics cards are more than half a bit better than NVIDIA's 40 series.
In terms of design, PCIe devices can communicate directly without going through the CPU, which means that in some high-performance computing scenarios, the graphics card GPU can directly exchange data between the storage nvmeSSD and the high-speed network card. Data does not need to be processed by the CPU, because scientific computing tasks also rely on the GPU At this time, the CPU has only become a "controller" and has been marginalized. Intel/AMD do not want to see this, so in the past they have deliberately set a limit on the bandwidth between the CPU and the PCIe device. The "bug" makes direct connection useless, and the way to get around this limitation is to add a PCIe bridge yourself.
But now it seems that due to too much pressure from Jiuzhou Technology, the once "technical difficulties" were instantly overcome and instead became a technological highlight.
There is also the AMD Software: Adrenalin Edition 22.3.1 software update released by AMD, which brings RSR super-resolution technology to players.
This technology not only brings amazing performance and stunning graphics in many game masterpieces, but also allows gamers to enjoy demanding visual features such as ray tracing.
Even the previous 6000 series 6700XT can rely on this technology to meet the balance between 4K image quality and frame number.
And under the ideal conditions in the laboratory, AMD's 7000 series card king has game performance comparable to that of Titan processor C. Even without considering the cost and heat dissipation, some hardware scores are even better. .
This made Lao Huang, who had previously vowed to draw a cake, instantly feel deflated.
The atmosphere in Daxia that is proud of Kyushu's technology has also begun to be affected to a certain extent by the promotion of certain forces.
After the mining disaster, those profiteers and second-rate dealers who did not become trapeze artists now feel more and more resentful when they see that Jiuzhou Technology is doing better.
(End of this chapter)